JPH1075051A - Visual inspection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the illuminance irregularity for the objects of inspection, in an visual inspection device. SOLUTION: Circular illumination light sources L1-L5, capable of adjustment of brightness are provided in tiers in vertical direction at the inside periphery face of a semicircular dome 7, and the brightness of the circular illumination sources L1-L5 in each stage can be adjusted and controlled by a probability distribution illumination producing circuit 6. A board 9 is arranged under the circular illumination light sources L1-L5, and the probability distribution illumination producing circuit 6 adjusts the brightness of circular illumination light sources at each stage, so that the brightness of the circular illumination light sources at the specified stage may be higher and that the brightness of the circular illumination light sources at the periphery of its specified stage. Under this illumination, the image of the solder fillets 11 on the board 9 are picked up by an image pickup device 3. Hereafter, a computer 5 switches the angle of illumination to the board by changing the specified stage high in brightness. At each of this switching, the said image pickup device 3 picks up the image of the object of inspection.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a visual inspection device, and more particularly to a visual inspection device suitable for visual inspection of a soldered portion on a mounting board of an electronic component.

[0002]

2. Description of the Related Art Conventionally, various devices have been proposed for optically inspecting the appearance of a soldered portion of a mounting board for electronic components. In these apparatuses, illumination is applied to a soldering portion to be inspected, the image is captured by an imaging device such as a CCD camera, and the appearance is determined by a computer based on the captured image. .

[0003] When these conventional visual inspection devices are classified,
It can be roughly classified into the following five types. 1) A device in which a one-stage illumination device is arranged so as to surround an inspection target, performs annular illumination, and determines the number of regions based on a captured image.

[0004] 2) An upper and lower two-stage illuminating device is annularly arranged so as to surround the inspection object, and the illuminating device of each stage is switched and illuminated. That takes the difference between the two and determines the difference.

[0005] 3) A multi-stage lighting device is arranged in a ring shape so as to surround the inspection object, the switching lighting of the lighting device of each stage is performed, and the area of the image captured by the lighting of each stage is determined. What to do.

[0006] 4) Multi-stage lighting devices are arranged in a ring shape so as to surround the inspection object, all the lighting devices are collectively illuminated, and the area is determined based on the captured image.

[0007] 5) A multi-stage lighting device is annularly arranged so as to surround the inspection object, and the switching lighting of the lighting device of each stage is performed. Performs dimensional matrix judgment.

[0008]

However, the technique 1) has a problem that the soldering state cannot be distinguished. An example thereof will be described with reference to FIGS.

FIGS. 64 (a) and 65 (a) are side views showing a state of soldering to an electronic component.
(B) and FIG. 65 (b) are image data captured from above through the imaging device, and a hatched portion represents a region (hereinafter, referred to as an image) W having a reflectance equal to or higher than a predetermined value. (Hereinafter, the same applies to the related art and the embodiment of the image data).

In FIG. 64 (a), the electrode 52 of the electronic component and the solder 53 are joined, and in FIG. 65 (a), the electrode 52 and the solder 53 are not joined.
Then, the image data captured by the image pickup apparatus in the state of both FIGS. (A) is exactly the same as shown in FIGS. 64 (b) and 65 (b). Therefore, in such a state, the determination cannot be made, and a defective soldering cannot be detected.

The technique 2) has a drawback that it is susceptible to noise because the determination method is based on the difference value. As long as the image is taken by the imaging device, the difference value does not become 0 when noise is mixed even if the same thing is intended to be taken. For example, as shown in FIG. 66 (a), when there is no soldering, the result of the two shootings may be divided as shown in FIGS. 66 (b) and (c). FIG. 66D shows image data after the difference between the image of FIG. 66B and the image of FIG. 66C is obtained. Noise due to shading unevenness is of a kind that has a very wide range of occurrence, and occurs suddenly. Although there is a correction method, it is a method of rounding an error, and it is impossible to make unevenness uniform. Since the difference value increases as the area becomes larger, the difference value may easily exceed the threshold value, and there is no method for preventing the difference value. Therefore, the inspection method based on the difference is incomplete as a technique of the solder inspection method.

The techniques 3) and 4) eliminate the problem 1) by irradiating the inspection object with multi-stage illumination, so that the same image is not obtained. In addition, when the area value is calculated, binarization processing is performed. However, since the luminance range in which shading unevenness occurs and the luminance range due to illumination light are significantly different, shading unevenness is cut off during this processing, and the problem of 2) noise occurs. Has been eliminated.

However, when making a determination based on the area value,
There is a problem shown in FIGS. FIG. 67A shows a case where the corner of the electrode 52 of the electronic component is round, and FIG.
(A) shows the case where the solder is bonded to the electrode 52 of the electronic component, but the image by the low-position illumination light is shifted to the edge side of the electrode 52.

FIGS. 67 (b) and 68 (b) show the image data of FIGS. 67 (a) and 68 (a), respectively. Since there is a slight difference in the position where the image W appears in FIGS. 67 (b) and 68 (b), it can be theoretically determined. However, in practice, it is difficult to determine the boundary of the determination because a mechanical error occurs in mounting the electronic component. Therefore, it is necessary to tighten the inspection standards so that unsoldered products do not accidentally pass through the inspection line. For this reason, the recognition rate is reduced as a whole.

FIG. 69 (a) shows a state where the solder 53 is joined to the electrode 52 of the electronic component, and FIG. 69 (b)
Indicates that the solder 53 is not soldered to the electrode 52. When the above two states are imaged,
If binarization is performed to determine the area, there is a problem that there is almost no difference between the two images. Although there is a slight difference in theory, it is not easy to discriminate between the two due to noise actually occurring. Further, when the solder is joined to the upper part of the electrode 52 of the electronic component in FIG. 69 (a), it may be determined that the solder is over-soldering. If it does, it must be determined as a solder defect. Therefore, if the soldering state shown in FIG. 69 (a) is not determined to be a solder defect, there is a possibility that the yield of the product by inspection lacks accuracy.

In the technique 5), pattern matching is performed, and when performing two-dimensional matrix determination, the number of conditions increases, and if all cases are covered, the image processing program becomes enormous. Therefore, it is practically impossible to cover all cases, and there is a problem that erroneous recognition easily occurs due to missing check.

In the prior art as described above, 2),
The techniques 3) and 5) switch the annular illumination light source to another stage of the annular illumination light source by turning on and off. For example, in the technique disclosed in Japanese Patent Application Laid-Open No. 5-303627, the quality of the soldered portion is inspected by sequentially switching and irradiating the annular multi-stage illumination, and taking an image with an imaging camera for each switching.

However, when illumination is performed by switching the light source between on and off, there is a problem that illuminance unevenness easily occurs on the inspection object due to an afterimage effect. The present invention has been made in order to solve the above problems, and an object of the present invention is to provide a visual inspection device capable of reducing uneven illuminance of an inspection object.

Another object of the present invention is to provide an appearance inspection apparatus which can simplify the processing of image data and can inspect the appearance of an inspection object with high accuracy.

[0020]

In order to solve the above problems, the invention according to claim 1 comprises an inspection light source having a plurality of annular illumination light sources whose brightness can be adjusted, and an annular illumination light source of each stage. A light source control unit for adjusting and controlling the luminance; an imaging unit for imaging an inspection object arranged under the annular illumination light source; and the luminance of a predetermined stage of the annular illumination light source among the plurality of annular illumination light sources by the light source control unit. An illumination control means for controlling the switching of the irradiation angle to the inspection object to another stage by changing the predetermined stage in a state where the luminance of the annular illumination light source around the predetermined stage is lowered. The device is the gist.

The gist of the invention of claim 2 is that, in claim 1, the light source control means adjusts the luminance distribution of the plurality of annular illumination light sources to be a probability distribution.

According to a third aspect of the present invention, there is provided the first or second aspect.
In, reflected light distribution obtaining means for obtaining reflected light distribution data of the inspection object based on the imaging data obtained from the imaging means, by switching the irradiation angle to the inspection object to another stage,
A center of gravity calculating means for calculating the center of gravity of the reflected light distribution area in each reflected light distribution data based on each of the obtained reflected light distribution data; and a barycentric position in each reflected light distribution data obtained by the center of gravity calculating means. A center-of-gravity movement trajectory determining means for determining a center-of-gravity movement trajectory; a storage means for storing standard data of the center-of-gravity movement trajectory; a standard data stored by the storage means; And an appearance determination unit that determines the appearance of the object to be inspected.

According to a fourth aspect of the present invention, in the appearance inspection apparatus according to the third aspect, the irradiation control means illuminates the inspection object with a light from a light source at a low angle with respect to the inspection object among the inspection light sources. It is intended to include a flat detecting means for measuring the amount of reflected light by irradiating the object and detecting the flatness of the inspection object based on the amount of reflected light.

The invention of claim 5 is the invention of claim 1 or claim 2.
In, reflected light distribution obtaining means for obtaining reflected light distribution data of the inspection object based on the imaging data obtained from the imaging means, by switching the irradiation angle to the inspection object to another stage,
A center of gravity calculating means for calculating the center of gravity of the reflected light distribution area in each reflected light distribution data based on each of the obtained reflected light distribution data; and a barycentric position in each reflected light distribution data obtained by the center of gravity calculating means. A center-of-gravity movement trajectory determining means for determining a center-of-gravity movement trajectory; and a solder bridge presence / absence determination means for determining presence / absence of a solder bridge of the inspection object based on the center-of-gravity movement trajectory data determined by the center-of-gravity movement trajectory determination means Is the gist.

According to the first and second aspects of the present invention, the inspection object is placed under the inspection light source, and in this state, the light source control means first increases the luminance of the annular illumination light source of the predetermined stage. The brightness of the annular illumination light source in each stage is adjusted so that the brightness of the annular illumination light source in the periphery of the predetermined stage is reduced, and an imaging object is imaged by the imaging means under this illumination. Hereinafter, the irradiation control means controls the switching of the irradiation angle with respect to the inspection object by changing the predetermined stage having a high luminance. For each switching control,
The imaging unit captures an image of the inspection target. Further, the light source control means adjusts the luminance distribution of the plurality of annular illumination light sources so as to be a probability distribution.

The present invention differs from the prior art in that there is no problem due to the multistage annular illumination. The conventional multi-stage annular illumination method is a method in which the annular illumination light source of each stage is turned on / off to perform switching illumination. However, in this method, the luminous intensity of the illumination significantly changes at the boundary. For example, when inspecting the soldering portion of the mounting board, as shown in FIG. 70A, the oxide film 54 on the surface of the soldering portion (solder fillet) 53 is formed. Due to the state and delicate unevenness of the blow hole 55 and the like as shown in FIG. 71 (a), image data as shown in FIG. 70 (b) or FIG.
There is a problem that the bright spot 56 due to one reflected light is divided into two. Note that the right side of FIG.
Is image data in the case where is not divided. In such a case, even if it is desired to treat that part as one area, the bright spot 5
When the image data taken in according to the reflectance (luminance) in No. 6 is divided into regions, it is split into two.

On the other hand, in the illumination method of the annular illumination light source having a plurality of stages according to the present invention, the luminous intensity of the annular light source at a predetermined stage is set as a peak, and the luminous intensity is reduced at a base within a certain reference range from the peak point. This illuminates the annular illumination light source of another stage according to the probability distribution function. When light of such a probability distribution type illumination intensity is applied to the inspection object, no light and dark occurs in subtle irregularities, and intense light actually interferes with each other in the subtle irregularities. And imaged. Therefore, in the present invention, illuminance unevenness does not occur.

One method of image processing includes a smoothing filter. Although it is possible to fill such a fine uneven portion with this smoothing filter, there is a problem that a portion to be actually observed is crushed. or,
There is also a method using a diffuser to diffuse the illumination light,
When a diffuser is used, light is spread over a wide area, and the recognition rate is deteriorated in subsequent processing, for example, in recognition processing.

In the probability distribution type illumination according to the present invention, light is not diffused (actually, there is some diffusion, but light from light sources in other stages adjacent to each other interferes with each other.
Diffusion is suppressed), and the illumination light can be concentrated on the inspection object.

Further, in the conventional multistage annular illumination method, the surface of the inspection object is inspected microscopically by switching illumination. Therefore, it is strongly affected by the unevenness of the reflected light due to the fine irregularities on the surface of the inspection object.
When trying to recognize the shape of such a microscopic inspection object (for example, a soldered portion), it is necessary to precisely set conditions such as where the concave is and how long the concave is. In the present invention, since the shape recognition is performed macroscopically, such condition setting is not necessary, and the shape recognition can be easily performed.

According to the third aspect of the present invention, the reflected light distribution obtaining means obtains reflected light distribution data of the inspection object based on the image data obtained from the imaging means. The center of gravity calculating means switches the irradiation angle with respect to the inspection object to another stage,
The center of gravity of the reflected light distribution area in each reflected light distribution data is calculated based on each of the obtained reflected light distribution data. The center of gravity movement trajectory determining means determines the center of gravity movement trajectory based on the position of the center of gravity in each reflected light distribution data obtained by the center of gravity calculation means. The appearance determining unit compares the standard data stored in the storage unit with the center-of-gravity moving trajectory data determined by the center-of-gravity moving trajectory determining unit to determine the appearance of the inspection object.

According to a fourth aspect of the present invention, the external appearance of the inspection object is determined by the same configuration as the third aspect of the present invention, and the flatness detection means controls the irradiation control means to control the light source of the inspection light source. The inspection object is irradiated with a light source from a low angle, and the amount of reflected light is measured. Then, the flatness of the inspection object is detected based on the reflected light amount.

According to the fifth aspect of the present invention, the reflected light distribution obtaining means obtains reflected light distribution data of the inspection object based on the image data obtained from the imaging means. The center of gravity calculation means,
The irradiation angle with respect to the inspection object is switched to another stage, and the center of gravity of the reflected light distribution area in each reflected light distribution data is calculated based on the obtained reflected light distribution data. The center of gravity movement trajectory determination means determines the center of gravity movement trajectory based on the position of the center of gravity in each reflected light distribution data obtained by the center of gravity calculation means. The solder bridge presence / absence determination means determines the presence / absence of a solder bridge of the inspection object based on the center-of-gravity movement trajectory data determined by the center-of-gravity movement trajectory determination means.

[0034]

BEST MODE FOR CARRYING OUT THE INVENTION

(First Embodiment) An embodiment of the present invention will be described below with reference to FIGS.

FIG. 1A is an overall schematic configuration diagram of an appearance inspection apparatus (hereinafter, simply referred to as “apparatus”) 1 of the present embodiment. This apparatus 1 is mainly composed of an illumination device 2, an imaging device 3, an image processing device 4, a computer 5, and a probability distribution illumination generation circuit 6 controlled by the computer 5, and performs an appearance inspection of a soldered portion. Things.

The illumination device 2 comprises a dome 7 having a hemispherical shape,
The dome 7 includes a plurality of LEDs 8 provided on the inner peripheral surface. In this embodiment, the plurality of LEDs 8 are annularly arranged at equal intervals in each of the first to fifth stages so as to be five stages vertically, and the first stage annular illumination light source L1 to the fifth stage annular The illumination light source L5 is used. Equal intervals are desirable in terms of treating light from all light sources equally, but they are not necessarily equal. The LED 8 can be adjusted not only by simple on / off control but also stepwise from a minimum luminance Lmin to a maximum luminance Lmax.

An opening is provided in the upper center of the dome 7, and the imaging device 3 is arranged in the opening. In the present embodiment, the imaging device 3 is constituted by a CCD camera. The imaging device 3 is connected to a computer 5 via an image processing device 4. Also,
The computer 5 controls the probability distribution illumination generation circuit 6 to adjust the luminance of the LEDs 8 at the respective stages to illuminate.

In FIG. 1A, the dome 7 is shown in cross section, and the substrate 9 has electrodes 10 of an electronic component and soldered portions (hereinafter referred to as solder fillets) 1 on its upper surface.
1 are shown in a state where they are joined.

Here, the correspondence between the area on the substrate 9 to be image-processed and the image data to be image-processed in the image captured by the imaging device 3 will be described with reference to FIG. FIG.
2B shows a plan view of a cross section of the solder fillet 11 shown in FIG. 2A, and a predetermined frame D indicated by a dotted line in FIG. Less than,
Simply referred to as “frame D”). Note that the image range on the substrate 9 actually captured by the imaging device 3 is wider than the frame D. In addition, as shown in the image FIG.
Part of 0 is also included.

The image in the frame D among the images captured by the imaging device 3 is
M × n (1 ≦ p ≦ m, 1 ≦ q ≦
n) is converted into matrix image dot data,
From the image processing device 4 onward, the image is processed as dot data. Incidentally, the (1, 1) dot shown in FIG. 2C is dot data corresponding to the upper left corner A of the frame D.

In FIG. 2B, K indicates an inspection line. The inspection line K is a data line of one dot width set on the matrix image dot data.
Not set above.

In FIG. 2B, Q indicates a predetermined width on the inspection line K. Also, t indicates a coordinate axis in the direction of the right end with the left end of the frame D as the origin, and t1 indicates the position coordinates of the end face of the electrode 10 in the frame D. The direction of the coordinate axis of t is equal to the direction of the coordinate axis of q shown in FIG.

Next, the probability distribution illumination generation circuit 6 is shown in FIG.
This will be described with reference to FIG. The I / O port 13 as an output port of the computer 5 has D / A channels 0 to D / A
Channel 8 (hereinafter referred to as “CH0” to “CH8”)
Are connected, and the output of each of CH0 to CH8 is connected to the non-inverting input terminal of the voltage follower circuit 14. The output terminal of each voltage follower circuit 14 is connected to the base of the transistor TR via a resistor R1. The collector of each transistor TR is connected to a DC power supply (48 V) via a series circuit consisting of 24 LEDs 8 each. In FIG. 3, the LED 8 is given a 2-digit or 3-digit suffix for each channel. The first digit of the two- or three-digit suffix indicates a channel, and the other digits indicate the individual numbers of the LEDs 8 in that channel. The emitter of each transistor TR is connected to a resistor R2.
Grounded. Each connection point between the emitter and the resistor R2 is connected to the inverting input terminal of each voltage follower circuit 14.

The probability distribution illumination generation circuit 6 is provided with a D / A common (COMMON) port 15 for each channel, and each D / A common port 15 is provided with a voltage follower circuit 14.
Is connected to the adjustment resistor R4 provided in the first stage. A resistor R3 is connected between the output terminal of each D / A channel and each D / A common port 15, and the output terminal of each D / A common port 15 is grounded. The offset of each voltage follower circuit 14 is adjusted in advance by the adjustment resistor R4.

FIG. 4 shows the arrangement of the LEDs 8 in each stage. That is, the first stage (the number of stages in the figure is 1)
The first stage annular illumination light source L1 is composed of LEDs 8 (L01 to L024) corresponding to CH0. The second stage (the number of stages in the figure) is CH1
And LED 8 corresponding to CH2 (L11 to L124, L21 to
L224) are interleaved and arranged, and these LEDs
8, a second-stage annular illumination light source L2 is formed. Third
In the stage (the number of stages is 3 in the figure), LEDs 8 (L31 to L324, L41 to L424) corresponding to CH3 and CH4 are alternately arranged and arranged, and these LEDs 8 constitute a third stage annular illumination light source L3. The fourth tier (the number of tiers in the figure is four) includes LEDs 8 (L51 to L5) corresponding to CH5 and CH6.
L524, L61 to L624) are alternately arranged and arranged,
These LEDs 8 constitute a fourth-stage annular illumination light source L4. The fifth stage (five stages in the figure) is the LED 8 (L71 to L724, L81 to L82) corresponding to CH7 and CH8.
4) are arranged alternately, and a fifth-stage annular illumination light source L5 is constituted by these LEDs 8.

Therefore, for example, when the first-stage annular illumination light source L1 is caused to emit light at a predetermined luminance, the computer 5
A 0 is designated and a predetermined digital signal (a digital signal corresponding to a predetermined analog voltage, the same applies hereinafter) is input thereto. Then, CH0 converts the digital signal into a predetermined analog voltage corresponding to the digital signal and inputs the same to the non-inverting input terminal of the corresponding voltage follower circuit 14. The voltage follower circuit 14 accurately applies the input analog voltage to the base of the transistor TR. Then, a predetermined collector current corresponding to the voltage applied to the base flows through the transistor TR. That is, the predetermined current corresponding to the predetermined digital signal input to CH0 is
(L01-L024), and these LEDs 8 emit light according to the amount of current.

For example, when the second-stage annular illumination light source L2 emits light at a predetermined luminance, the computer 5 addresses CH1 and CH2 and inputs a digital signal to them. As a result, the corresponding voltage follower circuit 14 applies a predetermined base voltage to the corresponding transistor TR,
The current corresponding to the applied predetermined base voltage is an LED.
8 (L11 to L124, L21 to L224), and these LEDs 8 emit light according to the amount of current. Similarly, in the other stages, when a predetermined digital signal is input to the corresponding D / A channel, a current corresponding to the digital signal flows through each corresponding LED 8,
The LEDs 8 at each stage emit light with a predetermined luminance according to the amount of current.

Here, the computer 5 operates the LE of each stage.
A calculation method and the like for calculating a voltage value applied to each channel of the probability distribution illumination generation circuit 6 from the I / O port 13 in order to emit D8 with a predetermined luminance will be described. Although the voltage value actually applied to each channel from the I / O port 13 of the computer 5 is a digital value,
Hereinafter, for convenience, the description will be made as a value converted into an analog voltage value.

FIG. 5 shows the relationship among the substrate 9 to be inspected, the light source (LED), and the imaging center axis M of the imaging device 3. A portion of the substrate 9 passing through the imaging center axis M is defined as a base point O, and any light source P (that is, L
The angle between the optical axis from the ED 8 and the plane of the substrate 9 is defined as θ. Given a certain angle θTYP with respect to the optical axis of a certain light source P, the brightness of each LED 8 at each stage is given in this embodiment so as to form a Poisson distribution as a probability distribution centered at the angle θTYP.
In the drawing, θc is an angle of a physical limit value for preventing the dome 7 and the imaging device 3 from interfering with each other, and θ0 is an initial value of θ, that is, an electronic component mounted on the mounting board 9 to be inspected. This is an angle that does not create shadows.

FIG. 6 shows a light emission luminance distribution when the luminance of the light source (LED) having the optical axis of the angle θTYP is peaked and another light source is along the Poisson distribution curve.
Then, the computer 5 calculates an applied voltage (analog converted value, hereinafter the same) for each channel of the probability distribution illumination generation circuit 6 by the following equation.

Vouti = Vconi × Ci where Vouti is i-stage illumination (i = 1, 2, 3, 4, 5).
Output voltage value (analog conversion value) of the computer 5 for
Vconi is a theoretical voltage value derived from the theoretical luminance value of the i-th stage illumination, and Ci is a correction value thereof.

The theoretical voltage value Vconi at each stage is obtained from a table of theoretical luminance values obtained based on the Poisson distribution. The method for obtaining this table is as follows. FIG. 7 shows a Poisson distribution curve as a probability distribution curve with luminance on the vertical axis and distribution distance on the horizontal axis. In the figure, the peak value is set to LTABLE1, and the brightness levels LTABLE2, LTABLE3, LTABLE4, and LABLE according to the Poisson distribution are respectively corresponding to equidistant positions around the peak value.
TABLE5 is defined. In this embodiment,
The distribution distances of the other brightness levels LTABLE around the peak value are set to be at equal intervals. However, the present invention is not limited to this. Good.

The voltage values corresponding to the luminance levels LTABLE1 to LTABLE5 obtained as described above, that is, the voltage values VTABLE1 to VTABLE5 () for obtaining the collector currents for the LEDs to emit light at the luminances of the levels LTABLE1 to LTABLE5. The theoretical value) is stored in advance in the RO as a storage means of the computer 5.
It is stored in M5a.

Here, for example, when making the third stage the brightest, the computer 5 selects the table value VTABLE1 as the theoretical voltage value Vcon3 of the third stage annular illumination light source L3. Then, the computer 5 selects the table value VTABLE2 as the theoretical voltage values Vcon2 and Vcon4 of the second-stage annular illumination light source L2 and the fourth-stage annular illumination light source L4, and the first-stage annular illumination light source L1 and the fifth-stage annular illumination. The table value VTABLE3 is selected as the theoretical voltage values Vcon1 and Vcon5 of the light source L5. For example, when making the fifth-stage annular illumination light source L5 brightest, the computer 5 selects a table value VTABLE1 as the theoretical voltage value Vcon5 of the fifth-stage annular illumination light source L5. Then, the computer 5 selects the table value VTABLE2 as the theoretical voltage value Vcon4 of the fourth-stage annular illumination light source L4, and selects the table VTABLE3 as the theoretical voltage value Vcon3 of the third-stage annular illumination light source L3. Further, a table value VTA is set as a theoretical voltage value Vcon2 of the second-stage annular illumination light source L2.
BLE4 is selected, and the theoretical voltage value V of the first-stage annular illumination light source L1 is selected.
The table value VTABLE5 is selected as con1.

In the present embodiment, as described above, when the computer 5 selects the brightest theoretical voltage value in a predetermined stage, the luminance sequentially decreases for the stages separated from the predetermined stage. The theoretical voltage value according to the Poisson distribution is selected as described above.

Next, a method of calculating the correction value Ci will be described. This correction value Ci is obtained in advance by a test. FIG.
Shows a diagram for explaining a test method for calculating the correction value Ci. Here, the case where the correction value C3 for the third-stage annular illumination light source L3 is calculated will be described. In FIG. 8, the spherical object 20 is placed on the imaging center axis M of the imaging device 3, and only the third-stage annular illumination light source L3 has the table value V
Light is emitted based on TABLE3, and the other annular illumination light sources are turned off. In this state, the image is taken by the imaging device 3, the image processing device 4 performs predetermined image processing, and then inputs image data to the computer 5. As described above, the image data is composed of a set of pixels arranged in a matrix of m × n pixels in the vertical and horizontal directions.

FIG. 9 shows the image data input by the computer 5 in a visualized manner.
The non-hatched portion indicates a portion B reflected by the LED (hereinafter, referred to as “reflected light data portion B”). In the figure, the computer 5 divides an area of the reflected light data portion B having a high reflectance, and samples luminance values x31, x32, x33,... X310 in order from the one having the highest reflectance.
Then, the sum X3 of the luminance values is obtained from the following equation.

[0058]

(Equation 1) Similarly, only the annular illumination light source of each stage emits light based on the table value VTABLE3, and the other annular illumination light sources are turned off. Based on the obtained image data, the sum X1 of the luminance values is obtained.
, X2, X4, X5.

Then, from the sums X1 to X5 of the luminance values obtained as described above, the maximum value XMAX is obtained by setting XMAX = max (X1, X2, X3, X4, X5).

By substituting the maximum value XMAX obtained as described above and the sums X1 to X5 of the luminance values in the image data when the annular light source in each stage emits light, the correction value Ci in each stage is obtained. Is calculated.

C1 = X1 / XMAX C2 = X2 / XMAX C3 = X3 / XMAX C4 = X4 / XMAX C5 = X5 / XMAX Next, the appearance inspection method of the solder fillet 11 by the apparatus 1 configured as above will be described. This will be described with reference to FIGS.

FIG. 10 shows an inspection processing routine executed by the computer 5 in the apparatus 1 according to the present embodiment. In step 10, the mounting substrate 9 is set, and after the setting is completed, in step 20, the angle counter θ is set to θ0. In step 30, the LED having the peak luminance of the lowermost annular light source (the fifth annular light source L5 in this embodiment) closest to the angle θ0.
The light emission pattern is called from the ROM 5a of the computer 5 and emits light based on the light emission pattern.

In this embodiment, there are five types of light emission patterns stored in the ROM 5a, and the light emission pattern is set as a fifth pattern, and the angle counter θ is set to θ0 + θ.
The fourth pattern at step, the third pattern at θ0 + 2θstep, the second pattern at θ0 + 3θstep, θ
In the case of 0 + 4θ step, the first pattern is read out, and the LED 8 emits light based on this LED light emission pattern. The arrangement of the LED 8, the imaging device 3, and the substrate 9 in each stage is as shown in FIG. As shown in the figure, θstep is a base point O on the substrate 9.
Represents the angle between the optical axes of the annular illumination light sources L1 to L5 and the adjacent optical axes, and has a constant angular width. Therefore, the angles between the optical axes are set equal to each other. Then, based on each pattern, the computer 5
Are applied to the probability distribution illumination generation circuit 6 (analog converted value) as follows. In the following equation, Vou
t1 to Vout5 are the first to fifth stage annular illumination light sources L, respectively.
These correspond to 1 to L5.

When the fifth (lowest) annular illumination light source L5 has the brightest fifth pattern, Vout1 = VTABLE5.times.C1 Vout2 = VTABLE4.times.C2 Vout3 = VTABLE3.times.C3 Vout4 = VTABLE2.times.C4 Vout5 = VTABLE1.times.C5 When the four-stage annular illumination light source L4 has the brightest fourth pattern, Vout1 = VTABLE4.times.C1 Vout2 = VTABLE3.times.C2 Vout3 = VTABLE2.times.C3 Vout4 = VTABLE1.times.C4 Vout5 = VTABLE2.times.C5 The third-stage annular illumination light source L3 is the brightest. In the case of the bright third pattern, Vout1 = VTABLE3.times.C1 Vout2 = VTABLE2.times.C2 Vout3 = VTABLE1.times.C3 Vout4 = VTABLE2.times.C4 Vout5 = VTABLE3.times.C5 If the second stage annular illumination light source L2 is the brightest second pattern, Vout1 = VTABLE2.times.C1 Vout2 = VTABLE1.times.C2 Vout3 = VTABLE2.times.C3 Vout4 = VTABLE3.times.C4 Vout5 = VTABLE4.times.C5 The first stage (uppermost stage) annular illumination light source L1 is the brightest. For one pattern becomes Vout1 = VTABLE1 × C1 Vout2 = VTABLE2 × C2 Vout3 = VTABLE3 × C3 Vout4 = VTABLE4 × C4 Vout5 = VTABLE5 × C5.

When step 30 is first entered, the fifth
The pattern is read, and based on this fifth pattern,
Computer 5 for the channel corresponding to each annular light source
Apply the applied voltage. As a result, the annular illumination light sources L1 to L5 at each stage emit light based on this pattern. That is, the fifth stage annular illumination light source L5, which is the lowermost stage of the illumination device 2, emits the brightest light, and the other stages emit light according to the Poisson distribution.

In the following step S40, an image from the image pickup device 3 which picks up an image of the solder fillet 11 of the mounting substrate 9 in a state where the fifth-stage annular illumination L5 emits the brightest light is captured. The area where the image processing is performed in the captured image is the area indicated by the frame D (FIG. 2).

Next, in step 50, various "filter processes" are performed on the image data. This “filter processing” will be described with reference to FIG. In step 51 of FIG. 11, an average value filtering process is performed.
This average value filtering process is a process of removing a level portion below the average value of the acquired data. Step 5
In 2, the smoothing counter value j is reset to 0. In step 53, a smoothing filter process is performed. Smoothing filter processing is processing to apply a small amount of smoothing so as not to change the shape of the mountain of image data. For example,
When the waveform in FIG. 12A is subjected to the smoothing process, the waveform in FIG.
The result is as shown in FIG. When this process ends, the smoothing counter value j is incremented and the routine goes to step 54.

In step 54, it is determined whether or not the smoothing counter value j is less than a specified number J. If the value is less than the specified number J, the process returns to step 53. Therefore, steps 53 and 54 are performed until the specified number of smoothing processes are performed.
Is repeated.

In step 54, when the smoothing counter value j has reached the specified number of times J, the flow shifts to step 55 to perform luminance normalization. This luminance normalization normalizes the value of the brightness (luminance) of the pixel in the range of 0 to 255, and A / D
This is a process performed to reduce errors in data processing after conversion. FIG. 13A shows imaging data for each fixed unit time before the luminance normalization processing. Due to the luminance normalization, the waveform between the broken lines in FIG. 13A becomes as shown in FIG. . When the brightness normalization process is completed, the process exits from the filter process. The image data that has been subjected to this filtering is hereinafter referred to as reflected light distribution data [I].

Returning to the flowchart of FIG. 10, the description will be continued. Hereinafter, the processing of the reflected light distribution data [I] in the present embodiment is performed only on the image data on the inspection line K shown in FIG. 2B among the image data composed of m × n pixels. Things.

From the above-mentioned “filter processing” to step 60
Then, in the same step, “region division” of the luminance data of the pixels of the image data surrounded by the frame D shown in FIG. This “region division” is performed using the minimum value of the luminance as a boundary. Note that this brightness is based on the solder fillet 1
It is proportional to the intensity of light reflected from 1. In other words, the portion having the minimum value is a portion of the curved surface (including the flat surface) of the solder fillet 11 where the shape change is small.

For example, FIG. 14A is a side view showing a state where the electronic component 10 is soldered to the mounting board 9, and FIG. 14B is an image of the solder fillet 11 of the mounting board 9. In the image data, a luminance curve (data) at pixels arranged on the inspection line K is shown. And
FIG. 15 shows the luminance curve (data) in FIG. 14B.
Is divided into regions, and in the example of the figure, the coordinate positions t2,
At t3, a local minimum exists, and the coordinate is divided at the coordinate position.

When the "area division" is performed in step 60, the computer 5 stores the number N of the divided areas (hereinafter referred to as "number of divisions") in a predetermined storage area of the RAM 5b.

Here, the number of divisions N corresponds to one obtained by adding one to the number of inflection points of the sectional shape of the inspection object. For example,
In the example of FIG. 17A, the solder fillet 11 has a monotonous concave shape, and in FIG. 17C, the solder fillet 11 has a monotonous convex shape, and both have an inflection point of zero. Therefore, the division number N is 1. FIG.
In the example of FIG. 7B, the unevenness is combined, and the number of inflection points is 1, that is, the number of divisions N is 2. In the example of FIG. 17D, the concave and convex portions are continuously combined, and the number of inflection points is 2, that is, the number of divisions N is 3.

Next, in "calculation of the center of gravity" of step 70, the coordinates of the center of gravity of the luminance distribution are calculated for each of the regions divided by the "region division". FIG. 16 is an enlarged view of the area 1 in FIG. 15 obtained by dividing the area, and a method of calculating the barycentric coordinate value G will be described based on this figure.

The center-of-gravity coordinate values G are calculated independently from each other in the divided areas by calculating the balance of the moment about T as shown in FIG. The result is obtained as the coordinate value of T which minimizes the result. That is, the difference P between the left and right moments about T is calculated, and the coordinate position T on the inspection line K at which the difference P is minimum is the barycentric coordinate value G. Here, L (t) is the luminance, and t1 is the end face coordinates of the electrode 10 of the electronic component as described above.

[0077]

(Equation 2) When the calculation of the barycentric coordinate value in each area is completed, the computer 5 temporarily stores the barycentric coordinate value of each area obtained from the image data in the RAM 5b when the illumination is performed in the fifth pattern, and In step 80, the angle count θ is incremented. That is, θstep is newly added to the previous angle count value θ. In the following step 90, it is determined whether or not the angle count value θ is smaller than θc. Angle count value θ is θ
If less than c, this determination is “YES” and the process returns to step 30.

Accordingly, in step 30, since the angle count value θ is incremented to be θ0 + θstep, the computer 5 reads the fourth pattern as a light emission pattern, and based on the fourth pattern, A predetermined voltage is applied to a channel corresponding to the annular light source. As a result, the annular illumination light sources L1 to L5 in each stage emit light based on this light emission pattern. That is, the fourth stage annular illumination light source L4 of the illumination device 2 emits the brightest light, and the other stages emit light according to the Poisson distribution.

The following steps 40 to 80 are processed in the same manner to determine the barycentric coordinate value of each area in the illumination state of the Poisson distribution in the fourth pattern, and store the barycentric coordinate value in the RAM 5b. In the following step 90, "YE
S ”is determined, and the process returns to step 30.

The computer 5 performs the same processing until the angle count value θ reaches θc in step 90. During this time, the coordinates of the center of gravity of each area in the illumination state of the Poisson distribution based on the third pattern, the second pattern, and the first pattern are stored in the RAM 5b.

At step 90, the angle count value θ
Is greater than or equal to θc, the determination is “NO”, and step 100
In step 100, a recognition sequence {Xk} based on the pixels on the inspection line K is created for the image data that has been inspected.

Hereinafter, creation of the recognition sequence {Xk} will be described. Here, the angle count value θ is the angle θ0 to θ
The z total barycentric coordinate values stored in the RAM 5b in the range of c are arranged in ascending order of coordinate values regardless of the angles θ0 to θc.

The barycentric coordinate values thus obtained are represented by G 1, G 2
,... Gz (G1 <G2 <... <Gz). Then G1
, G2,..., Gz are determined as θ1, θ2,..., Θz, and are compared with the barycentric coordinate values G1, G2,.

Next, Xk = (. Theta.k-.theta.0) /. Theta.step (k = 1, 2,... Z) is defined, and a recognition sequence corresponding to the above-mentioned irradiation angles is as follows.

Coordinate values G1, G2, G3,..., Gk,..., Gz Irradiation angle θ1, θ2, θ3,..., Θk, θz Recognition sequence X1, X2, X3,. 10 °, θ0 = 30 ° and θc = 7
Assuming 0 ° and five illumination stages, the irradiation angle θ is 30
°, 40 °, 50 °, 60 ° and 70 °. At this time, for example, θ1 = 30 °, θ2 = 50 °, θ3 = 40
If θ, θ4 = 30 °, θ5 = 60 °, θ6 = 70 °,..., The recognition sequence {Xk} is as follows in accordance with the above defined equation.

The coordinates of the center of gravity G 1, G 2, G 3, G 4, G 5, G 6,..., The irradiation angles 30 °, 50 °, 40 °, 30 °, 60 °, 70 °,..., The recognition sequence 0, 2, 1, 0, 3, 4,... Next, the “group” of the recognition sequence {Xk} will be described. For example, in the above illumination configuration, the recognition sequence {Xk}
Is {0, 1, 2, 3, 4, 4, 3, 2, 1, 0}.

In this sequence, all the numbers are arranged in ascending or descending order, that is, (0, 1, 2, 3, 4)
And (4,3,2,1,0) are defined as a "group". Then, the number of “groups” existing in the sequence is defined as the number S of groups in the recognition sequence {Xk}. In this example,
The number of groups S is two.

Further, this “group” is given a group number s in order from the left. In this example, "group"
(0, 1, 2, 3, 4) is the group number 1 and “group” (4, 3, 2, 1, 0) is the group number 2. (Hereinafter, the group of group number 1 is described as group 1) The meaning of this "group" will be described below.

According to the definition, the numbers in the recognition sequence {Xk} can correspond to the annular illumination light sources L1 to L5. That is, the annular illumination light source “L5” is set to “0” in the recognition sequence, “L4” is set to “1”, “L3” is set to “2”, “L2” is set to “3”, and “L1” is set similarly. "4" can be corresponded. That is, the fact that the numbers in the recognition sequence {Xk} are arranged in the ascending or descending order (the existence of a “group”) means that the barycentric coordinates move in the switching order of the annular illumination light sources L1 to L5. That is, the presence of “group” in the recognition sequence {Xk} means that the shape of the solder fillet 11 has regularity. For example, "ascending group" (0, 1, 2, 3, 4)
17A, the concave shape shown in FIG. 17A is obtained, and if the “descending group” (4, 3, 2, 1, 0) is present, FIG.
The convex shape shown in (c) is recognized. In the present embodiment, various shapes of the solder fillet 11 are recognized based on the regularity of the recognition sequence {Xk}.

The computer 5 recognizes the recognition sequence {X}
k} is created, and the created recognition sequence {Xk} and the barycentric coordinate values related to the numerical values constituting the recognition sequence {Xk} are both stored in the RAM 5b, and the process proceeds to step 110. In step 110, a "collation with master pattern" routine is executed, and when the processing is completed, the inspection processing routine for the soldered portion is terminated.

Next, the "collation with master pattern" routine will be described in detail with reference to the flowcharts shown in FIGS. 18 to 21 and FIGS. 22 to 26. When this routine is entered, in step 200, a group number counter (s) for counting the group number s is set to 1, and in step 201, data reduction for the group s is performed. That is, of the barycentric coordinate values in the group of group number s, the maximum value is GMAXs and the minimum value is GMAX.
Assuming MINs, a difference GRANGEs between the maximum value GMAXs and the minimum value GMINs is calculated.

In step 202, the difference GRANGEs
Is determined to be equal to or less than a reference value Q for determining whether or not the data actually indicates the presence of the solder fillet 11.
The reference value Q is a predetermined width on the inspection line K, and is a predetermined value determined by the solder material, the type and shape of the electronic component, and the like. In step 202, if the difference GRANGEs is equal to or less than the reference value Q, it is determined to be “YES”, and the process proceeds to step 209.
The intrinsic judgment flag Bs is set to 1, and the routine jumps to step 208. That is, it is determined that the data of the solder fillet 11 includes noise data such as a foreign substance and an oxide film. In step 202, the difference GRANGEs
Is greater than the reference value Q, the determination is "NO", the routine proceeds to step 203, the intrinsic judgment flag Bs is reset to 0, and the routine proceeds to step 204. That is, it is determined that the solder fillet 11 exists without fail.

In step 204, it is determined whether or not the minimum barycentric coordinate value GMINs is equal to or smaller than a determination value t1. This is to determine whether or not there is a minimum barycentric coordinate value GMINs on the electrode 10 of the electronic component. It should be noted that the judgment value t1 is the same as that in FIG.
As shown in (b), when the left end face of the frame D is set to 0, the position coordinates of the end face of the electrode 10 of the electronic component in the frame D are shown.

In step 204, if the minimum barycentric coordinate value GMINs exceeds the determination value t1, "NO" is determined, and the routine proceeds to step 210, where the barycentric attribute flag A
s is reset to 0, and the routine proceeds to step 206. It is determined that at least the minimum barycentric coordinate value GMINs is not on the electrode 10.

If it is determined in step 204 that the minimum barycentric coordinate value GMINs is equal to or smaller than the determination value t1, the determination is "YES".
s is set to 1 and the routine proceeds to step 206. That is, it is determined that the minimum barycentric coordinate value GMINs is on the electrode 10 and the solder is on the electrode 10.

When the process proceeds from step 205 or step 210 to step 206, in step 206,
It is determined whether the recognized number sequence of the group with the group number s is an ascending sequence.

Here, the correspondence between the ascending order and the descending order of the recognition sequence {Xk} and the shape of the solder fillet will be described. For example, FIG.
The sequence of recognition of the shape of the solder fillet shown in FIG.
{Xk} = {0, 1, 2, 3, 4, 4, 3, 2, 1,
It consists of 0｝. The column of (0, 1, 2, 3, 4) in the first half constitutes group 1, and the column of (4, 3, 2, 1, 0) in the latter half constitutes group 2. In group 1, “0” is the minimum coordinate value GMIN1 and “4” is the maximum coordinate value G in the sequence.
MAX1. In group 2, in the sequence,
"4" is the minimum coordinate value GMIN2, "0" is the maximum coordinate value GMAX2
belongs to.

Therefore, according to this example, group 1 is determined to be in ascending order of "01234". Group 2
In this case, the order is “43210” in descending order. FIG. 22 (b) shows the plane of the electrode 10, and in the frame D, the left arrow indicates the descending order, and the right arrow indicates the ascending order. Move in the ascending or descending direction (the direction of the arrow in the figure).

As another example, in the case of the solder fillet 11 shown in FIG. 23A, {Xk} = {0,
1,2,3,4,4,3,2,1,0,0,1,2,2
3, 4｝. The column of (0, 1, 2, 3, 4) in the first half constitutes group 1 (ascending order), and the middle (4, 3, 2, 2)
The column of (1, 0) constitutes group 2 (descending order), and the column of (0, 1, 2, 3, 4) in the latter half constitutes group 3 (ascending order).

In the group 1, "0" is the minimum coordinate value GMIN1 and "4" is the maximum coordinate value GMAX1 in the sequence. In the group 2, “4” is the minimum coordinate value GMIN2 and “0” is the maximum coordinate value GMAX2 in the sequence. or,
In group 3, “0” is the minimum coordinate value GMIN3 in the sequence,
“4” is the maximum coordinate value GMAX3. FIG. 23 (b)
Indicates a plane of the electrode 10, and in a frame D,
A leftward arrow indicates a descending order, and a rightward arrow indicates an ascending order, and the barycentric coordinate value similarly moves in this direction.

As described above, at step 206, it is determined whether or not the sequence of the group s is in ascending order. If it is determined that the sequence is ascending, at step 207, the ascending attribute flag C is determined.
s is set to 1 and the process proceeds to step 208. If it is determined in step 206 that the numerical sequence of the group having the group number s is not ascending but descending, the ascending attribute flag Cs is reset to 0 in step 211, and the process proceeds to step 208.

When the process proceeds from step 207, step 211 or step 209 to step 208, it is determined whether or not the group number s is less than the group number S. If the group number s is less than the group number S, it is determined that the setting or resetting of the attribute flag of another group has not been completed yet (step 21).
Then, the process proceeds to step 201 where the group number counter is incremented.

When the process of setting or resetting the center-of-gravity attribute flag As, the intrinsic judgment flag Bs, and the ascending attribute flag Cs is completed for all groups, step 2 is executed.
At 08, the group number s is equal to the group number S, so the determination in step 208 is “NO”, and the routine proceeds to step 213 (FIG. 19).

In step 213, the groove number counter is set to 1, and the flow shifts to step 214 to determine whether or not the intrinsic judgment flag Bs is set to "1". When the intrinsic judgment flag Bs has been reset to 0, this step 214 is judged as “NO”,
Jump to step 217. In step 214, when the intrinsic judgment flag Bs is set to "1", it is determined that the data of the solder fillet 11 includes noise, so that step 214 is determined as "YES", and the process proceeds to step 215. .

In step 215, since it is determined that the group currently being processed contains noise, the data containing this noise is deleted, and the number of the group after deletion is reduced by one for all groups. Update.

For example, FIG. 24A shows a matrix of the center-of-gravity attribute flag As, the intrinsic determination flag Bs, and the ascending attribute flag Cs in the group of each group number s (s is 1 to 4 in this example). It is. Group 2
Is set to "1", it is determined that group 2 contains noise. Therefore, the column α related to the group 2 is deleted. After deletion, the next column group (in this example, group 3) is newly set as group 2, and the next column group (in this example, group 4) is newly set as group 3 (FIG. 24). (B)).

In the next step 216, step 215
Since the group related to the noise of the above has been deleted, the number of groups S and the group number s are reduced by one, and
In S17, it is determined whether or not the currently processed group number s (the updated number performed in step 215) is less than the number S of groups. In step 217,
If the group number s is less than the group number S, it is determined that there is still a group to be processed, the process proceeds to step 218, and the group number counter is incremented.
Return to step 214.

Therefore, the processing of steps 214 to 217 is repeated until this processing is completed for all groups. In step 217, when these processes are completed for all groups and the group number counter becomes equal to the group number S, step 2
The determination in step 17 is “NO”, and step 219 (FIG. 20)
Move to

The reason why the processing routines of steps 213 to 218 are provided is as follows. For example,
In FIG. 17B, when the data of the solder fillet 11 does not include noise, the solder fillet 1
Assume that the recognition sequence for 1 is as follows.

{X k} = {0, 1, 2, 3, 4, 4,
However, if the surface of the solder fillet 11 has a blowhole 20 or a foreign substance such as dust or an oxide as shown by a dashed line, the recognition sequence for the solder fillet 11 is as follows, for example. It may be.

{Xk} = {0, 1, 2, 3, 4, 2,
1,0,4,3,2,1,0} Therefore, the processing routine of the above-mentioned steps 213 to 218 is provided in order to remove the noise relating to the blowhole 20.

Next, a description will be given of the steps after step 219. At step 219, the group number counter is set to 1
And the process proceeds to step 220. Step 22
0, the group of group number s and the group number (s
+1) It is determined whether or not there is a group in which the Euler number is 1 among the groups. That is, it is determined whether there is a hole between both groups. The number of Eulers is detected by a processing routine (not shown) different from the soldering inspection processing routine. Since the detection of the Euler number is publicly known, a detailed description thereof is omitted. However, in the entire image data, pixels having the same luminance are connected to form a ring-shaped connected component. Is a low portion, that is, when it is a hole, the Euler number is set to 1.

In step 220, the Euler number is 1
If not, the determination in step 220 is “NO”, and in step 228, the Euler number attribute flag E
Reset s to zero. If the Euler number is 1 in step 220, the determination in step 220 is “YE
S ", and in step 221, the Euler number attribute flag Es is set to 1. In step 222, it is determined whether or not the currently processed group number s is less than the group number S. If the group number s is smaller than the group number S in step 222, it is determined that there is still a group to be processed, the process proceeds to step 229, and the group number counter is incremented.
Return to

Therefore, steps 220, 221, 228, and 22 are performed until this processing is completed for all groups.
Steps 2 and 229 are repeated. In step 222, these processes are completed for all groups,
When the groove number counter becomes equal to the group number S, the determination in step 222 is “NO”, and the process proceeds to step 223.

The processing routine of steps 219 to 222, 228 and 229 is provided for the following reason. As shown in FIG. 25, when reflected light having a ring-shaped (closed loop) hatched portion is detected in the range of the region D (corresponding to the range of image data) as shown in FIG. So shaped solder fillet 1
1 is in an over-soldering state. Therefore, in the above step, the Euler number is detected by another routine, and when the Euler number is 1, the ring-shaped portion becomes a connected component, and a hole exists in the connected component.
Accordingly, in step 220, the group number s
If there is a hole between the group of (a) and the group of the group number (s + 1), this can be determined as the appearance shape in the over-soldered state.

Next, in step 223, it is determined whether or not the number of groups S is 0. If the number of groups is 0, the process proceeds to step 230, in which non-soldering determination is performed,
Exit this processing routine. Note that the number of groups S is 0
This means that even if the illumination of the annular illumination light sources L1 to L5 is switched to the annular illumination light source whose luminance is peaked in the Poisson distribution, the position of the center of gravity did not move in ascending or descending order. The fact that the position does not move means that there is no curved surface, that is, the surface is flat. Therefore, it is determined that there is no solder.

If the number of groups S is not 0 in step 223, the process shifts to step 224 to determine whether or not the number of groups S is 1. If the group number S is not 1, it is determined that the solder fillet 11 to be inspected does not have a simple shape, and the process jumps to step 233 (FIG. 21).

If the number of groups S is 1 in step 224, it is determined that the appearance of the solder fillet 11 to be inspected is a simple shape, that is, concave or convex, and the process proceeds to step 225.

In step 225, it is determined whether or not the Euler number attribute flag E1 (= Es) of the group (group 1) whose group number is 1 is 1. If the Euler number attribute flag E1 is 1, the step 231 is executed.
Then, an error determination is made, and this processing routine ends.

This processing is determined as an error for the following reason. That is, in step 224, the number of groups S is determined to be one. On the other hand, in another processing routine, since the number of Eulers is detected to be one, the Euler number attribute flag E1 is set to one in step 221.
Is set to If the detection that the Euler number is 1 is correct, the shape of the solder fillet 11 is a mountain shape, and the number of groups should be at least two. Therefore, since the detection result of the Euler number and the group number do not match, this is determined as an error.

At the step 225, the group 1
If the Euler number attribute flag E1 (= Es) is not 1, that is, if it is 0, the process proceeds to step 226 to determine whether the ascending attribute flag C1 (= Cs) is 1.
If the ascending attribute flag C1 is 1, step 23
2 shows the appearance of the solder fillet 11 shown in FIG.
Normally, it is determined that the shape is concave as shown in FIG. 7A, and this processing routine ends. On the other hand, even when the ascending attribute flag C1 is 0, in step 227, the external shape of the solder fillet 11 is changed as shown in FIG.
It is normally determined that the shape is a convex shape as shown in (c), and this processing routine ends.

Next, the case where the process jumps to step 233 (FIG. 21) will be described. Moving to step 233,
First, it is determined whether the group number S is less than a reference value. This reference value is used to limit the maximum number of groups S, and the number of groups larger than this is determined by experiments as an impossible value. Therefore, if the value exceeds the reference value, the process proceeds to step 237 to determine an error, and the processing routine is terminated. If the group number S is less than the reference value, the process proceeds to step 234.

In step 234, it is determined whether or not the Euler number attribute flag E1 is 1 and the ascending attribute flag C1 is 1 for the groups having the group number S of 2 or more. That is, in this determination step, it is determined whether or not the solder fillet 11 has a mountain shape having a ring-shaped reflected light and the movement of the center of gravity is in ascending order.

In this step 234, the Euler number attribute flag E1 is 1 and the ascending order attribute flag C1
Is 1, the flow shifts to step 235 to determine whether or not the gravity center presence / absence attribute flag A1 is 0. That is,
It is determined whether or not the group 1 is located on the electrode 10 of the electronic component. Center of gravity attribute flag A1
Is 0, the determination in step 235 is made “YES”, and the process proceeds to step 238, where the group 1 is not on the upper surface of the electrode 10 and is shown, for example, in the solder fillet 11 in FIG. Since it has such a chevron shape, it is determined that solder wetting is defective, and this processing routine ends.

If it is determined in step 235 that the gravity center presence / absence attribute flag A1 is not 0 but is 1, then step 23 is executed.
The determination of 5 is “NO”, and the process proceeds to step 236,
The group of group number 1 is on the upper surface of the electrode 10 and, for example, the solder fillet 11 shown in FIG.
Since it has a chevron shape as shown in (1), it is determined that the solder is excessive, and this processing routine ends.

At step 239 after shifting from step 234, the Euler number attribute flag E1 is 1 for group 1, and the ascending attribute flag C1 is 0 for group 1.
Is determined. That is, in this determination step, it is determined whether or not the solder fillet 11 has a mountain shape having a ring-shaped reflected light and the movement of the center of gravity is in descending order.

If the Euler number attribute flag E1 is 1 and the ascending order attribute flag C1 is 0, the flow shifts to step 240 to determine whether or not the gravity center presence / absence attribute flag A1 is 0. That is, group 1 is the electrode 1 of the electronic component.
It is determined whether or not it is located above zero.

When the center-of-gravity attribute flag A1 is 0,
The determination in step 240 is made “YES”, and step 24
In step 2, the group 1 is not on the upper surface of the electrode 10 and the solder fillet 11 has a chevron shape. Therefore, it is determined that the solder is over-soldered, and the processing routine ends.

If it is determined in step 240 that the barycenter presence / absence attribute flag A1 is 1 instead of 0, the process proceeds to step 24.
The determination of 0 is "NO", and the process proceeds to step 241.
Since this group 1 is on the upper surface of the electrode 10 and has an upwardly extending icicle ridge shape as shown, for example, in the solder fillet 11 of FIG. End the routine.

In the step 239, the group 1
, The Euler number attribute flag E1 is 1, and
If the ascending attribute flag C1 does not satisfy both conditions of 0, the process proceeds to step 243. Here, the solder fillet 11 does not form a chevron shape, for example, as shown in FIG.
(B), a normality determination is made assuming that the external shape as shown in FIG. 17 (d) has an S-shape, and the processing routine ends.

The effects obtained by the above-described embodiment will be described below. In the present embodiment, each group of the recognition sequence has a center-of-gravity attribute flag As, an intrinsic determination flag Bs, an ascending attribute flag Cs, and an Euler number determination flag Es, and a combination of the attribute flag and the determination flag , The shape of the solder fillet 11 was recognized. The criteria for setting and resetting the attribute flag are all determined based on the data relating to the barycentric coordinate values. Then, by dividing the movement locus of the barycentric coordinate value into groups each having a certain group, and determining whether the group unit is an ascending or descending sequence, the external shape can be grasped.
Since the calculation of the barycentric coordinate value is performed by a simple calculation,
Despite the simple technique, the appearance shape of the solder fillet 11 can be reliably recognized.

In the present embodiment, a Poisson distribution is adopted as the probability distribution, one of the annular illumination light sources L1 to L5 is illuminated to have a peak luminous intensity, and the luminous intensity is calculated from the peak point according to the Poisson distribution function. The annular illumination light source of the other stage was illuminated so that the image quality was reduced. As a result, when light having a Poisson distribution-type illumination intensity is applied to the mounting substrate 9, no light or darkness occurs in the subtle irregularities, and intense light actually interferes with the subtle irregularities. The image is integrated. Therefore, in the present embodiment, the influence of the uneven illuminance and the uneven gloss of the surface of the solder fillet 11 due to the annular illumination light sources L1 to L5 is reduced, so that more accurate image data can be obtained as the imaging data of the solder fillet 11. Become.

In this embodiment, as described in step 215 of the routine shown in FIGS. 18 to 21, when noise is included in the group data, the group is determined based on the intrinsic determination flag. It was removed so that the remaining groups could recognize the external shape. As a result, the external shape of the solder fillet 11 can be accurately recognized without being disturbed by noise data such as foreign substances and oxide films.

(Second Embodiment) Next, a second embodiment will be described with reference to FIGS.
In this embodiment, a description will be given mainly of points different from the first embodiment.

In this embodiment, each of the annular light sources L1 to L5 is illuminated in accordance with the Poisson distribution and the predetermined steps are switched so as to have a peak luminous intensity, similarly to the above-described embodiment. The sectional shape of the solder fillet 11 is drawn using the coordinates of the center of gravity (moving locus).

FIG. 27 shows the substrate 9 not soldered.
FIG. 27 shows a sectional view of the inspection area D (FIG.
(See (b)), the length of the inspection line K is L, and the length from the end face of the electrode 10 of the electronic component to the right end of the inspection area D shown in FIG. 27 is l.

FIG. 28 shows a drawing control routine executed by the computer 5. This drawing control routine is executed using the data obtained by the appearance inspection processing routine after the appearance inspection processing routine of the first embodiment is completed. Then, based on the drawing control routine, the computer 5 draws the sectional shape of the solder fillet 11 on a display (not shown).

[0138] Of the barycentric coordinate values G obtained by the visual inspection processing routine, those of (L-1) ≤G≤L are used as effective barycentric coordinate values G'k used in this drawing control routine.
And Before entering the drawing control routine, the computer 5 sorts the effective center-of-gravity coordinate values G'k in ascending order and renumbers the effective center-of-gravity coordinate values G'k to values 1 to D (D≤z).

The sequence of the effective barycentric coordinate values G'k is as follows. The sequence of effective centroids'G'1,G'2, ..., G'k, ...,
G′D｝ In addition, when the irradiation angle θ when the barycentric coordinate value of G′k is obtained is represented by a sequence corresponding to this sequence of effective centroids, the following is obtained.

{Θ1, θ2,..., Θk,..., ΘD} Also, the slope (ta
n θk) is known in advance from experimental values and the like, and assuming that the slope of this interpolation line is a, the following is obtained.

{A1, a2,..., Ak,... AD} The irradiation angle θ and the inclination a of the interpolation line are fixed values because the annular illumination light sources L1 to L5 in the illumination device 2 are not movable. These values are stored in the ROM 5a of the computer in advance.

For example, when the irradiation angles of the annular illumination light sources L1 to L5 are in the following sequence, the sequence of the slope of the interpolation curve is as follows. Irradiation angle sequence: {20, 30, 40, 50, 60} Sequence of inclination of interpolation straight line: {tan35 °, tan30 °, tan25
°, tan20 °. tan15 ° Now, the control routine is entered. In step 300, the storage area of the RAM 5b for storing the following various data is initialized as follows.

Q = 0, b * = 0, p = L, b = 0, k = D Here, q and b * are the end points of the interpolation straight line (referred to as interpolation points).
Represents the x coordinate and the y coordinate. Similarly, p and b represent the x coordinate and y coordinate of the start point of the interpolation line. In this drawing control routine, drawing is started from the one having the largest effective barycentric coordinate value number, that is, the number D.

In step 310, it is determined whether or not the number of the effective barycentric coordinate value is k ≠ 1. If the effective barycentric coordinate value is k ≠ 1, “YES” is determined, and the routine proceeds to step 320. In step 320, the x coordinate value q of the interpolation point between the effective barycentric coordinate values G′k and G ′ (k−1) is calculated by the following equation, and the process proceeds to step 330.

Q = (G'k + G'k-1) / 2 If the number of the effective barycentric coordinate value is k = 1 in step 310, "NO" is determined, and step 36 is performed.
Then, the process proceeds to step 330, where the x coordinate value q = 0, and the process proceeds to step 330.

After calculating the x coordinate value q of the interpolation point in step 320 or 360, in step 330, the y coordinate value of the interpolation point between the effective barycentric coordinate values G'k and G '(k-1) b * is calculated by the following equation, and the process proceeds to step 340.

B * = ak × (p−q) + b The effective barycentric coordinate value G′k = (p, b). Accordingly, the computer 5 draws a straight line between the interpolation point (q, b *) obtained in steps 320 and 330 and the start point (p, b) of the interpolation straight line, and in step 350, assigns the number k to (k -1), the x coordinate p of the start point of the interpolation straight line is updated to q, and the y coordinate b is updated to b *.

At step 370, it is determined whether or not the number k ≠ 0. In step 370, if the number k ≠ 0, the process returns to step 310, and the process is continued in step 31.
Process steps 0 to 360.

By the processing of steps 310 to 360,
In step 370, when the number k = 0, the determination is “NO” and the routine ends. FIG. 29 shows an example in which the effective barycenter coordinate value is k = 3. In the figure, H indicates an interpolation point.

The effects obtained by the above-described embodiment will be described below. In the second embodiment, the cross-sectional structure of the outer shape of the solder fillet 11 can be known. Also, several inspection lines K are taken, and the solder fillet 11 is similarly formed.
By obtaining the cross section, the amount of solder in the solder fillet 11 can be calculated.

(Third Embodiment) In the first and second embodiments, the appearance and shape of the solder fillet 11 in the electrode of the electronic component with respect to the mounting board 9 were inspected. The purpose is to perform a general-purpose appearance inspection without targeting the solder fillet 11. In this embodiment, the same hardware configuration as in the first embodiment is provided.

FIG. 30A is a plan view of the inspection object 16 and FIG. 30B is a sectional view. Conventionally, in the case where the appearance of the inspection object 16 is as shown in FIGS. 30A and 30B, even if image processing is performed, a clear feature such as an edge does not easily appear in the image, and the image is light and dark. However, there is no way to cope with the conventional image processing.

That is, in the conventional image processing technique, the hole 17 as shown in FIG.
As shown in (b), a singular point such as a concave portion 18 having a corner was required. FIGS. 31A and 31B are plan views of the inspection object 16.

In the present embodiment, as shown in FIG. 32 (a), image data obtained by imaging with the imaging device 3 is set so that the inspection line K is not a straight line but corresponds to the shape of the inspection object 16. Alternatively, the inspection line K is determined according to the appearance required for the inspection. Therefore, in this embodiment, unlike the first embodiment in which the inspection line K is a straight line, the inspection line K is a curve, and among the pixels constituting the image data, A corresponding pixel is selected, and a visual inspection is performed based on the luminance on this pixel.

In FIG. 32 (b), a one-dimensional luminance curve of FIG. 32 (c) is obtained, for example, on the inspection line K as shown. Accordingly, when the appearance inspection processing routine is executed and processed in the same manner as in the first embodiment, the computer 5 performs processing such as area division based on the sampled data in the same manner as in the first embodiment, and performs area division. The center-of-gravity movement trajectory is obtained for each of the regions, and after obtaining the trajectory, the appearance of the inspection object 16 is inspected by comparing it with the master pattern. In this embodiment, in this embodiment, various shape data corresponding to the appearance of the inspection object 16, that is, center-of-gravity movement trajectory data corresponding to each shape is stored in the ROM 5 a in advance as a master pattern. Shall be.

The effects obtained by the above-described embodiment will be described below. In the third embodiment, the inspection line K is, for example, a plane circle as shown in FIG.
In the inspection object as shown in FIG. 34, or by crossing a plurality of inspection lines K with each other, the appearance inspection of various shapes can be performed.

If the drawing control routine is used in the same manner as in the second embodiment based on the respective barycentric coordinate values obtained in the third embodiment, the cross-sectional shape of the inspection object 16 can be determined. As shown in FIGS. 31C and 31D, the inspection target 16
It is also possible to draw a contour line R on the image data.

In the third embodiment, the appearance can be inspected and recognized even if the appearance does not have a peculiar shape. Therefore, in particular, when positioning a work (inspection object) having no peculiar shape, there is no need to provide a positioning part for image recognition processing having a peculiar shape and pattern in a part other than the inspection object. It is also possible to perform positioning using only the inspection object. (Fourth Embodiment) Next, a fourth embodiment will be described with reference to FIG.
This will be described with reference to FIGS.

In this embodiment, a solder bridge is provided between a solder fillet 11 of an electrode 10 of an electronic component soldered to a mounting board 9 and a solder fillet 11 of an electrode 10 of another adjacent electronic component. The present invention relates to a visual inspection device capable of inspecting whether or not it is formed.

FIG. 35A is a plan view of the solder fillet 11 of the electronic component mounted on the substrate 9 when there is no solder bridge, and FIG.
FIG. 35 (c) shows a land 1
The case where the silk screen printing 26 is performed between the two is shown.

In the conventional solder bridge inspection, a judgment is made based on the area of the reflected light, and if the area of the reflected light is large, it is assumed that there is a solder bridge. However, in the case of the prior art, if silk screen printing is performed between the component lands or if there is a step on the land 12, the illumination light hits the portion and the reflected light is imaged, so even if there is no solder bridge, There is a case where it is erroneously determined that there is a solder bridge.

In this embodiment, there is provided a visual inspection apparatus for detecting a solder bridge without the risk of such erroneous determination. Since the hardware configuration is the same as that of the first embodiment, The same components are denoted by the same reference numerals.

FIG. 36 is a plan view of the mounting board 9, FIG. 37 (a) is a sectional view of the solder bridge, and FIG. 37 (b) is a plan view thereof. FIG. 38 shows the imaging device 3 and
The arrangement relationship between the substrate 9 and the annular illumination light sources L1 to L5 is shown.

In this embodiment, the inspection line K is, as shown in FIG. 36, a land 12 where solder bridges are likely to occur.
The direction is set so as to extend across, that is, a direction orthogonal to the mounting direction A of the electronic component. In this embodiment, the width of the inspection line K is also one dot width of the pixel.

Also in this embodiment, FIG.
9 is executed. In this processing routine, the same processing as in the first embodiment is performed in steps 10 to 100. Therefore, for example, when one-dimensional luminance data is obtained as shown in FIG. 40 and there is a minimum value, region division is performed in step 60 as in the first embodiment.

In step 120, it is determined whether or not the "group" (moving of the center of gravity in ascending or descending order accompanying the switching of the annular illumination light source) described in the first embodiment exists.

That is, in this step, for each of the divided areas, whether the obtained recognition sequence {Xk} has a portion arranged in descending order or a portion arranged in ascending order is determined. Is determined.

For example, when a sequence of {Xk} = {4,3,2,1,0,1,3} is obtained in a certain region as described above, (4,3,2,1,0 ) Is in descending order. Also, {Xk} = {3,4,0,
If a sequence of 1,2,3,4} is obtained, (0,1,
2, 3, 4) are assumed to be in ascending order.

When there is a descending order or an ascending order in the recognition sequence, it is determined that a "group" exists. Then, in step 130, when it is determined in step 120 that a "group" exists, it is determined that there is a solder bridge. On the other hand, if it is determined in step 120 that no “group” exists, it is determined that there is no solder bridge. Then, this processing routine ends. Steps 120 and 130 constitute a solder bridge presence / absence determining means.

In the above-described embodiment, in FIG. 38, the luminance peaks are sequentially switched from the annular illumination light sources L5 to L1 as shown by arrows in FIG. 38, and the other annular illumination light sources follow the Poisson distribution. Was switched.

FIG. 41A shows a side view on the inspection line when there is no solder bridge, and FIG. 41B shows an image taken at that time. 42 to 45, each figure (a) shows a schematic cross-sectional view on an inspection line when the solder bridge 24 is present, and each figure (b) shows an image taken at that time. Here, in FIG. 42 (b) to FIG. 45 (b), the case where the annular illumination light source L5 emits light at the peak luminance is indicated by hatching rising to the right, and the annular illumination light source L5 is shown.
The case where 1 is emitted at the peak luminance is indicated by hatching to the lower right.

In FIG. 41, since there is no solder bridge, the image has no reflected light and no "group" (shift of the center of gravity) exists. In the case of the chevron-shaped solder bridge 24 in FIG. 42 (a), as shown in FIG. 42 (b), when the annular illumination light source L5 emits light at its peak, the reflected light that has been split into left and right light forms the annular illumination light source L1. When the light emitted at the peak, it gathered at the center.

The concave solder bridge 24 shown in FIG.
Also in the case of FIG. 43, as shown in FIG.
The reflected light, which was split right and left when the light emitted at the peak of 5, was collected at the center when emitted at the peak of the annular illumination light source L1.

In the case of the upward-sloping solder bridge 24 shown in FIG. 44 (a), as shown in FIG.
When the annular illumination light source L1 emitted light at the peak, it moved to the right.

In the case of the upward-sloping solder bridge 24 shown in FIG. 45 (a), as shown in FIG.
When the annular illumination light source L1 emitted light at the peak, it moved to the left.

As described above, the "group" (movement of the center of gravity) varies depending on the shape of the solder bridge, but in any case, the "group" exists. Accordingly, when there is no solder bridge as shown in FIG. 41, there is no “group”, and therefore, by judging the existence of this “group”, the solder bridge can be detected.

Conventionally, image data captured by an imaging device is binarized and then subjected to recognition processing. However, when the silk screen printing 26 is performed between the lands 12 as shown in FIG. 35C, or when the wiring pattern 27 is between the lands 12 as shown in FIG. -If the illumination is switched off, uneven lighting occurs. Therefore, especially when the wiring pattern 27 exists, when the image data is binarized, the image data may or may not remain in the image data.
For this reason, detection of a solder bridge may not be possible. However, in the present embodiment, since the binarization processing is not performed, a portion such as the wiring pattern 27 does not remain or does not remain, so that stable recognition processing can be performed, and the above-described problem occurs. There is no.

In the case of the conventional image recognition, when an image is taken between the lands 12 in FIG. 46, an image 27a of the wiring pattern as shown in FIG. 47 is captured. When the recognition process is performed based on the captured wiring pattern image 27a, it may be determined that there is a solder bridge even though the solder bridge does not exist for the image 27a.
However, in the present embodiment, such a case does not occur because the image is not subjected to the binarization processing.

The effects obtained by the above-described fourth embodiment will be described below. According to the fourth embodiment, the solder bridge 24 between the lands 12
Is reliably detected and silk screen printing 2
6 and the wiring pattern 27 are not mistaken for the solder bridge 24. (Fifth Embodiment) Next, a fifth embodiment will be described with reference to FIG.
This will be described with reference to FIGS. In this embodiment, since the hardware configuration has the same configuration as that of the first embodiment, the same configuration is denoted by the same reference numeral.

In this embodiment, as shown in FIG.
When the cross-sectional shape of the solder fillet is a right triangle, that is, when the illumination reflection surface of the solder fillet is flat (hereinafter, referred to as
Even the flat fillet 11a) can be recognized.

In the flat fillet 11a, since the curvature of the solder fillet hardly changes, the above-described shift in the center of gravity does not occur. Therefore, FIG.
Although the flat fillet 11a as shown in FIG. 10 may be determined to have a normal solder fillet shape, the flat fillet 1a is not used in the appearance inspection processing routine (FIG. 10).
1a is not determined to be normal.

Therefore, in this embodiment, a "flat fillet detection" process is added to the above-described appearance inspection process routine. FIG. 50 shows the appearance inspection processing routine. The "flat fillet detection" process performed as the process of step 15 of the visual inspection process routine shown in FIG. 50 will be described below with reference to FIG.

In step 151 of FIG. 51, two types of low-angle light sources among the annular illumination light sources L1 to L5, for example, as shown in FIG.
5 and individually irradiate the solder fillets to convert the reflected light into image data. Note that this image data is binarized into a high level “1” or a low level “0” by a predetermined threshold.

Here, the annular illumination light source is irradiated from a low angle because the shape of the solder fillet is flat.
This is because, as shown in FIG. 49B, the fillet 11a strongly reflects the illumination light from a low irradiation angle with a wide area Bf as shown in FIG. That is, in the present embodiment, the flat fillet 11a is detected using the reflected light pattern Bf unique to the flat fillet 11a.

In step 152, step 15
OR of the image data (binary data) obtained in step 1
R), and obtain the sum Uθ. Next step 15
In 3, the following inequality sign is determined.

Uθ> U × F ₃ (0 <F ₃ <1) where, Uθ: the sum of logically ORed pixel data U: the sum of all pixel data at high level “1” (hereinafter simply referred to as “1”) High-level sum S) F ₃ : A predetermined constant determined by an experiment (hereinafter simply referred to as a constant F ₃ ).

That is, in step 153, the sum U
If θ is larger than the sum of the high-level sum U and the constant F ₃ , it is determined that the reflection fill amount is larger than the predetermined value and the flat fillet 11a is detected. Then, in the next step 154, the appearance inspection processing routine ends as normality determination.

On the other hand, in step 153, the total sum Uθ
If is less than or equal multiplied by the constant F ₃ to the high-level sum U is the amount of reflected luminance is flat fillets 11a smaller than a predetermined value is judged not to have been detected.
Then, the flow shifts to step 20 in FIG. 50, where the appearance inspection processing routine described above is performed.

The effects obtained by the fifth embodiment described above will be described below. According to the fifth embodiment, it is possible to detect the flat fillet 11a in which the center of gravity does not move. (Sixth Embodiment) Next, a sixth embodiment will be described with reference to FIG.
This will be described with reference to FIGS. In this embodiment, since the hardware configuration has the same configuration as that of the first embodiment, the same configuration is denoted by the same reference numeral.

In this embodiment, as shown in FIG.
Small particles (hereinafter simply referred to as foreign particles) 28 such as solder residue, dust, protrusions due to solder plating, and lump of flux.
On the unsoldered land 12 or the solder fillet 1
In the case where it adheres to the surface 1, the influence of the reflected light from the foreign matter 28 is removed.

This "foreign substance detection" is performed as the processing of step 16 of the appearance inspection processing routine shown in FIG.
The processing will be described. First, all the annular illumination light sources L1 to L5
Is irradiated on the inspection object. Then, the reflected light is imaged to convert the luminance into data. Next, as shown in FIG. 52, the image data is projected in the X direction and the Y direction shown in FIG. 52, and the data widths of the projection are assumed to be A _X and A _Y , respectively. Further, as shown in FIG. 52, X-direction and Y-direction data width, respectively and L _X and L _Y. At this time, the ratios of the data widths A _X and A _Y by projection to the data widths L _X and L _Y are R _X and R _Y , respectively.
It is represented by the following equation.

R _X = A _X / L _X , R _Y = A _Y / L _Y At this time, T _RX and T _RY are used for R _X and R _Y , respectively.
A predetermined threshold value is set, and only when all of the following conditional expressions (6-1) R _X ≧ T _RX (6-2) R _Y ≧ T _RY are satisfied,
Alternatively, it is determined that the small foreign matter 28 does not exist on the solder fillet 11. Thereafter, the above-described visual inspection processing routine shown in step 20 and subsequent steps in FIG. 53 is performed.

On the other hand, the above conditional expressions (6-1) and (6-2)
If at least one of them is not satisfied, it is determined that the small foreign matter 28 exists on the unsoldered land 12 or on the solder fillet 11, and the mask processing is performed. In this masking process, the reflected light data on the foreign matter 28 of the small piece is removed from the pixel data, that is, the reflected light data on the foreign matter 28 of the same small piece is reset to “0”. Thereafter, the above-described visual inspection processing routine shown in step 20 and subsequent steps in FIG. 53 is similarly performed.

This "foreign matter detection" processing may be performed in the "filter processing" in step 50 shown in FIG. In this case, the "foreign matter detection" processing is performed before the luminance normalization in step 55 in the flowchart of the "filter processing" shown in FIG. 11, and is performed for each annular illumination light source.

The effects obtained by the above-described sixth embodiment will be described below. According to the sixth embodiment, even when the small foreign matter 28 adheres to the unsoldered land 12 or the solder fillet 11, the influence of the reflected light from the foreign matter 28 is reduced. Can be removed, and the accuracy of the appearance inspection is improved. (Seventh Embodiment) Next, a seventh embodiment will be described with reference to FIG.
This will be described with reference to FIGS. In this embodiment, since the hardware configuration has the same configuration as that of the first embodiment, the same configuration is denoted by the same reference numeral.

In this embodiment, the reflected light having a peculiar shape can be detected, and for example, the crescent-shaped reflected light Bm generated at the soldered portion of the lead 10a shown in FIG. Is what you can do.

When the lead 10a shown in FIG. 54 is soldered normally (hereinafter referred to as a lead solder fillet) 11b, the reflected light is reflected in FIG.
It shines clearly in a crescent shape as shown in FIG. In the case of no solder, the lead slightly reflects at the edge of the lead and becomes a U-shaped Bs as shown in FIG. 55 (b).

The appearance of the lead solder fillet 11b can be detected by increasing the resolution of the reflected light image in the above-described method of moving the center of gravity. However, the number of data pixels increases, and the processing (calculation) takes time. In short, the inspection takes a long time. Therefore, in this embodiment, in step 17 of the appearance inspection processing routine of this embodiment shown in FIG. 62, "singular shape detection" is performed by the fixed point sampling method to shorten the inspection time. Hereinafter, this “detection of a unique shape” will be described.

First, all the annular illumination light sources L1 to L5 are irradiated to the vicinity of the lead solder fillet 11b to be inspected. Then, the reflected light is imaged to convert the luminance into data. Next, as shown in FIG. 56, the imaging data is projected in the X and Y axis directions, respectively, and the position of the solder fillet 11b on the data is detected. That is, FIG.
The coordinates of X ₁ , X ₂ , Y ₁ and Y _{2 shown} in FIG. 6 are determined.

Next, the above X ₁ , X ₂ , Y ₁ and Y ₂
The area A as shown in FIG. 57 is determined based on the coordinates of
The brightness data of the area A is sampled (fixed point sampling method) to detect the presence or absence of solder at the lead portion.

FIG. 58 is an enlarged view of the area A shown in FIG.
In FIG. 58, M _X , M _Y , L _X and L _Y are predetermined lengths determined in advance according to the component sizes. Also, _UL , _UM1 and _UM2 are the sums of the luminances of the pixels in the area shown in FIG. Here, the shape of the bright spot in the area A is determined based on the following inequality sign. _{That, (7-1) (U M1 +} U M2) ≦ F M (7-2) U L ≦ F L ( where F _M, F _L is a constant determined by experiment) when satisfied either alone, region The bright spot shape of A is determined to be a crescent shape, and it is determined that the soldering of the lead 10a is normal.

On the other hand, the above conditional expressions (7-1) and (7-2)
If at least one of them is not satisfied, the bright spot shape of the region A is determined to be a U-shape, and the soldering of the lead 10a is determined to be unsoldered.

The results of these determinations are flagged in “detection of unusual shape” in step 17 shown in FIG. 62, and in the “collation with master pattern” process in step 110 shown in FIG. This flag is referred to when is performed. That is, even if the unsoldering judgment is made in the “matching with the master pattern”, if the flag indicating that the soldering is normal in the “detection of the unique shape” is raised, the soldering of the lead 10a is finally finished. Is determined to be normal.

By the way, the unique shape detected by this "single shape detection" process is not limited to the above crescent shape.
For example, as shown in FIG. 59, a V-shaped reflected light Bv formed at the time of soldering the lead 10a can also be detected. Hereinafter, detection of the V-shaped singular shape will be described.

First, similar to the above-described detection of the crescent-shaped peculiar shape, the lead solder fillet 11b to be inspected is used.
All the annular illumination light sources L1 to L5 are irradiated to the vicinity. Then, the reflected light is imaged to convert the luminance into data.

Next, the image data is projected in the Y-axis direction shown in FIG. 60B, and the average value of the projection is obtained.
Then, as shown in FIG. 60 (b), the area widths V _WA and V _WB for determining the areas (A), (B) and (C) shown in FIG. 61 from the intersection of the average value and the projection value. And _VWC . Vgap shown in FIG. 61 is a predetermined width of a non-sampling region defined for each inspection object.

Subsequently, FIG. 61 determined as described above
In the areas (A), (B) and (C) shown in FIG. 7, the bright pixel numbers V _A , V _B
And to count the V _C. Then, based on the following inequality conditions, detection and determination of a V-shaped singular shape are performed.

That is, (7-3) V _C (1-F ₁ ) <V _A <V _C (1 + F ₁ ) (7-4) V _B F ₂ <V _A (7-5) V _B F ₂ < V _C (where F ₁ and F ₂ are constants determined by experiments, and 0
Only when all three conditions of <F ₁ < _1, 1≪F ₂ ) are satisfied, the V-shaped singular shape is detected. Here, in the conditional expression (7-3), the condition is that there is no great difference between the number of bright points in the area (A) and the number of bright points in the area (C). The conditional expressions (7-4) and (7-5) are based on the condition that the number of bright spots in the area (B) is smaller than that in the areas (A) and (C).

That is, only when all of the above three conditions are satisfied, a V-shaped bright spot shape as shown in FIGS. 59 and 60A is detected. When the V-shaped bright spot shape is detected, it is determined that the soldering of the lead 10a is normal.

[0210] The effects obtained by the seventh embodiment described above will be described below. According to the seventh embodiment, detection of the lead solder fillet 11b, which requires high resolution in the method of moving the center of gravity and requires time for image processing, detects a singular bright spot shape such as a crescent shape and a V shape. By doing so, the time required for appearance inspection can be shortened.

The above embodiment can be embodied by changing the configuration as follows. In all of the above embodiments, the annular illumination light source has five stages L1 to L5, but is not limited thereto. For example, the number of stages of the annular illumination light source is four, six,
It may be composed of stages or eight stages. Further, the number of LEDs constituting one light source is not limited to 24 or 48.

In all the above embodiments, the probability distribution illumination is the Poisson distribution illumination. However, if the probability distribution illumination is along a probability distribution function having a peak point and a base within a certain reference range, It can be used instead of the Poisson distributed illumination. In addition, basically, it is only necessary to sequentially irradiate light by multi-stage illumination, and it is not always necessary to follow the probability distribution function.

In all of the above embodiments, the LED is used as the light source. Instead, an optical fiber for introducing a laser beam is inserted into the inner peripheral surface of the dome 7, and the inspection object is inserted from the end surface. You may make it irradiate with respect to it.

In all of the above embodiments, the irradiation order of the annular illumination light sources L1 to L5 is from L5 to L1, that is, from the bottom to the top, but may be from the top to the bottom. .

In all of the above embodiments, the equation shown in (Equation 2) is used for calculating the barycentric coordinate value, but the present invention is not limited to this. When the image quality is poor (when the brightness of the light is unclear), for example, when the background of the image is bright and the luminance level of the background image is 20% or more of the maximum luminance level, or when the image If there is unevenness, the following equation using the squared value L ² (t) instead of the pixel luminance L (t) in (Equation ² ):

[0216]

(Equation 3) May be used to calculate the barycentric coordinate value from the coordinate value t that minimizes equation ( ₂ ).

In the barycentric coordinate value obtained by using the equation P ₂ shown in (Equation 3), since the high luminance level is further emphasized, the pixel position having the highest luminance is statistically determined using the least squares method. The same effect as obtained by the analysis is obtained. That is,
Since the barycentric coordinate value is calculated by reducing the influence of unclear light and darkness, the barycentric coordinate value can be more accurately obtained even when the image quality is poor.

In all of the above embodiments, the filter processing is performed in step 50 of the appearance inspection processing routine, but the contents of the filter processing are not limited to those shown in FIG. For example, an “inertia filter” process may be added to the pixel luminance data L (t) on the inspection line K before the luminance normalization in step 55 shown in the filter processing routine of FIG. Good.

This inertial filter processing is expressed by the following equation. L (t) = {I (t-1) × I (t) × I (t + 1)}
/ 65,536 (provided that 0 ≦ I (t) ≦ 255) By this inertial filter processing, data noise N as shown in FIG. 63 (a) can be removed as shown in FIG. 63 (b). For example, it is possible to eliminate the influence of white spots (defects of the image pickup device) of the CCD camera 3 and the influence of shavings of the substrate attached to the solder fillet 11. In addition, since the intermediate image data value is shifted up and down by the inertial filter processing, the image data is sharpened. As a result, the accuracy of the appearance inspection of the inspection object is improved.

All of the above embodiments can be arbitrarily combined with each other, and it is of course possible to implement all of the above embodiments.
Furthermore, of the technical ideas grasped from the above-described embodiment, technical ideas other than the claims are described below together with effects.

1) A center-of-gravity movement trajectory calculating means for calculating a center-of-gravity movement trajectory range, an intrinsic judgment means for judging whether or not the center-of-gravity movement trajectory range is larger than a predetermined value, and a noise data 5. The visual inspection apparatus according to claim 3, further comprising a noise data removing unit that removes the noise data when it is determined that there is a noise data.

According to the configuration, even if noise data is taken in due to the influence of an oxide film or the like in the image data, the noise data is removed, so that other judgments are not adversely affected and a more accurate external appearance is achieved. Inspection becomes possible.

[0223]

According to the first and second aspects of the present invention, the illuminance unevenness of the inspection object can be reduced.

According to a third aspect of the present invention, the visual inspection is performed by obtaining the center of gravity of the reflected light distribution area in the obtained image data of a plurality of inspection objects, and calculating the shift of the center of gravity with respect to the center of gravity related to other image data. Can be performed easily and reliably.

According to the invention of claim 4, since the flatness of the illumination reflection surface of the inspection object can be determined, a solder fillet having a flat illumination reflection surface can be detected. The invention of claim 5 is
Since the presence or absence of unevenness of the inspection object can be determined from the movement locus of the center of gravity, the detection of the solder bridge can be easily performed.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram showing an embodiment in which a visual inspection device according to the present invention is applied to a soldering portion inspection device.

FIG. 2 is a schematic diagram showing a relationship between a soldered portion of a mounting board and image data to be processed.

FIG. 3 is a block diagram showing an electrical configuration of a probability distribution illumination generation circuit.

FIG. 4 is a schematic diagram showing an LED array of a lighting device.

FIG. 5 is a schematic diagram illustrating a relationship between a substrate to be inspected, a light source, and an imaging center axis M of the imaging device.

FIG. 6 is a graph showing a light emission mode of the lighting device according to the embodiment.

FIG. 7 is a graph showing an example of a table according to a Poisson distribution.

FIG. 8 is a schematic diagram of a test method for calculating a correction value Ci.

FIG. 9 is a schematic diagram showing an example of image data obtained by an imaging device to obtain a correction value Ci.

FIG. 10 is an exemplary flowchart illustrating the processing procedure of a captured image according to the embodiment;

FIG. 11 is a flowchart showing a filter processing procedure in a processing routine of the captured image.

FIG. 12 is a graph showing an averaging process in the filter process.

FIG. 13 is a graph showing a normalization processing mode in the filter processing.

FIG. 14 is a schematic diagram illustrating a relationship between a side surface of a solder fillet and a one-dimensional luminance curve based on an inspection line.

FIG. 15 is a graph showing a region division mode based on the luminance curve.

FIG. 16 is a graph showing an enlarged region 1 obtained by dividing the region.

FIG. 17 is a cross-sectional (side) view showing various solder fillet shapes.

FIG. 18 is a flowchart illustrating a procedure of a process for checking a master pattern in an inspection routine;

FIG. 19 is a flowchart showing a procedure of a matching process with a master pattern in an inspection routine.

FIG. 20 is a flowchart showing a procedure of a matching process with a master pattern in an inspection routine.

FIG. 21 is a flowchart showing a procedure of a collation process with a master pattern in an inspection routine.

FIG. 22 is a schematic diagram showing a relationship between a solder fillet shape and a group.

FIG. 23 is a schematic diagram showing a relationship between a solder fillet shape and a group.

FIG. 24 is a schematic diagram illustrating a mode of removing noise from an image data group.

FIG. 25 is a schematic view showing an example of a solder fillet in which the number of Eulers is 1;

FIG. 26 is a cross-sectional (side) view showing various solder fillet shapes.

FIG. 27 is a cross-sectional view of a substrate.

FIG. 28 is a flowchart illustrating a drawing control routine according to the second embodiment;

FIG. 29 is an explanatory view illustrating a solder fillet.

FIG. 30 is a plan view and a cross-sectional view of an inspection object.

FIG. 31 is an explanatory diagram of an inspection object.

FIG. 32 is an explanatory diagram showing a relationship between an inspection object, an inspection line, and a luminance curve.

FIG. 33 is an explanatory diagram showing a relationship between an inspection object and an inspection line.

FIG. 34 is an explanatory diagram showing a relationship between an inspection object and an inspection line.

FIG. 35 is a plan view showing solder fillets, solder bridges, and silk-screen printing on a substrate.

FIG. 36 is a plan view showing inspection lines in a fourth embodiment.

FIG. 37 is a cross-sectional view and a plan view of a solder bridge.

FIG. 38 shows an imaging device 3, a substrate 9, and an annular illumination light source L.
Explanatory drawing which shows arrangement | positioning relationship with 1-L5.

FIG. 39 is a flowchart illustrating an appearance inspection processing routine according to the fourth embodiment;

FIG. 40 is a graph showing a one-dimensional luminance curve.

FIG. 41 is an explanatory diagram in the case where there is no solder bridge on the inspection line K.

FIG. 42 is an explanatory diagram in the case where a chevron-shaped solder bridge is present on the inspection line K;

FIG. 43 is an explanatory diagram in the case where there is a concave solder bridge on the inspection line K;

FIG. 44 is an explanatory diagram in the case where there is a solder bridge of another shape on the inspection line K.

FIG. 45 is an explanatory diagram in the case where there is a solder bridge of another shape on the inspection line K.

FIG. 46 is a plan view of the inspection object.

FIG. 47 is an explanatory diagram showing an image obtained by conventional image processing.

FIG. 48 is a cross-sectional (side) view showing the shape of a flat solder fillet.

FIG. 49 is an explanatory diagram of flat solder fillet detection.

FIG. 50 is a flowchart showing a processing procedure of a captured image according to the fifth embodiment;

FIG. 51 is a flowchart showing a flat fillet detection processing procedure in the processing routine of the captured image;

FIG. 52 is an explanatory diagram showing the relationship between a foreign object and its projection.

FIG. 53 is a flowchart showing a processing procedure of a captured image according to the sixth embodiment;

FIG. 54 is a cross-sectional (side) view showing a solder fillet shape of a lead wire.

FIG. 55 is an explanatory diagram showing a reflected light form of a solder fillet of a lead wire.

FIG. 56 is an explanatory diagram showing the relationship between crescent-shaped reflected light and its projection.

FIG. 57 is an explanatory diagram of an inspection area A.

FIG. 58 is an enlarged view of a region A shown in FIG. 57;

FIG. 59 is an explanatory diagram showing V-shaped reflected light.

FIG. 60 is an explanatory diagram showing the relationship between V-shaped reflected light and its projection.

FIG. 61 is an explanatory diagram of an area division in an inspection area D.

FIG. 62 is a flowchart showing a processing procedure of a captured image according to the seventh embodiment;

FIG. 63 is a graph showing a processing mode of inertial filter processing.

FIG. 64 is an explanatory diagram showing a soldering state of an electronic component and a captured image thereof.

FIG. 65 is an explanatory diagram showing a soldering state of an electronic component and a captured image thereof.

FIG. 66 is an explanatory view showing an unsoldered state of an electronic component and a captured image thereof.

FIG. 67 is an explanatory diagram showing electrode shapes of an electronic component and a captured image thereof.

FIG. 68 is an explanatory diagram showing a soldering state of an electronic component and a captured image thereof.

FIG. 69 is an explanatory view showing a soldering state of an electronic component and a captured image thereof.

FIG. 70 is an explanatory view showing a soldering state of an electronic component and a captured image thereof.

FIG. 71 is an explanatory diagram showing a soldering state of an electronic component and a captured image thereof.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Appearance inspection apparatus, 2 ... Illumination apparatus, 3 ... Imaging device,
4 image processing device, 5 computer, 6 probability distribution illumination generation circuit, 7 dome, 8 LED, 9 substrate
Reference numeral 10: electrode, 11: solder fillet, 13: I / O port, 16: inspection object, 20: blow hole, 24:
Solder bridge, 26 ... Silk screen printing, 27 ...
Wiring pattern, L1 to L5 ... annular illumination light source.

Claims (5)

[Claims] 1. An inspection light source having a plurality of annular illumination light sources whose brightness can be adjusted, light source control means for adjusting and controlling the brightness of each stage of the annular illumination light source, and an inspection object arranged under the annular illumination light source Imaging means for imaging an object, by the light source control means, in a state in which the brightness of the annular illumination light source of a predetermined stage among the plurality of annular illumination light sources is high and the brightness of the annular illumination light source of the periphery of the predetermined stage is low, An illumination control means for switching and controlling an illumination angle with respect to the inspection object by changing a predetermined stage. 2. An appearance inspection apparatus according to claim 1, wherein said light source control means adjusts the luminance distribution of said plurality of annular illumination light sources to be a probability distribution. 3. A reflected light distribution obtaining means for obtaining reflected light distribution data of an inspection object based on image data obtained from an imaging means; A center of gravity calculating means for calculating the center of gravity of the reflected light distribution region in each reflected light distribution data based on the light distribution data; and a movement locus of the center of gravity based on the position of the center of gravity in each reflected light distribution data obtained by the center of gravity calculating means A center-of-gravity movement trajectory determining unit, a storage unit that stores standard data of a center-of-gravity movement trajectory, ,
The appearance inspection apparatus according to claim 1, further comprising: an appearance determination unit configured to determine an appearance of the inspection object. 4. An illumination control means for irradiating the inspection object with illumination light from a light source having a low angle with respect to the inspection object among the inspection light sources, measuring an amount of reflected light, and based on the amount of reflected light. 4. The visual inspection apparatus according to claim 3, further comprising flatness detection means for detecting the flatness of the inspection object. 5. A reflected light distribution obtaining means for obtaining reflected light distribution data of an inspection object based on image data obtained from an imaging means; A center of gravity calculating means for calculating the center of gravity of the reflected light distribution region in each reflected light distribution data based on the light distribution data; and a movement locus of the center of gravity based on the position of the center of gravity in each reflected light distribution data obtained by the center of gravity calculating means And a solder bridge presence / absence determining means for determining the presence or absence of a solder bridge of the inspection object based on the center of gravity movement trajectory data determined by the center of gravity movement trajectory determination means. Item 3. An appearance inspection apparatus according to Item 2.