JP3299397B2 - Image processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for performing edge processing on a multivalued image signal such as a signal obtained by reading an image on a document in a digital copying machine or a signal generated by a computer.

[0002]

2. Description of the Related Art In a digital copying machine or a printer, an image signal obtained by reading an image on a document by an image reading device of the digital copying machine, or an image signal generated inside a computer and input from the computer. Are subjected to necessary and appropriate edge processing.

For example, in a digital copying machine,
A halftone portion composed of a line portion or a character portion and a halftone dot of the input image which is an image on the document is identified, and the edge of the line portion or the character portion is sharpened so that the sharpness is enhanced.
In the halftone section, edge processing is performed on a multi-valued image signal input from the image reading device so that the edge is smoothed and the halftone is reproduced well.

In a printer which prints out an image from a computer, an unnatural step or jag is reduced particularly in a line portion or a character portion due to font expansion of an input image which is an image from the computer. Edge processing is performed on a multi-valued image signal input from a computer so as to obtain a smooth line or character.

Recently, as an image output system such as a digital copying machine or a printer, a copying function for reading and processing an image on a document and outputting the image on a recording medium, and a method for processing an image from a computer have been proposed. A composite system having a printer function for outputting on a recording medium and a function as a facsimile has been considered.

In such a composite system, conventionally, a flag is added to an input multi-valued image signal to distinguish whether the input image is an image read by an image reading device or an image from a computer. In the edge processing, the input multi-valued image signal is separated into an image signal from the image reading device and an image signal from the computer by the flag, and the image signal from the image reading device is connected to the line portion as described above. In addition, so-called adaptive edge processing that distinguishes between a character portion and a halftone portion is performed, and for an image signal from a computer, edge processing that smooths lines and characters by font expansion as described above as necessary. Is to be made.

FIG. 15 shows an example of a conventional image processing apparatus for edge processing in the composite system. An input multi-valued image signal Di to which the above-described flag is added is supplied to an image separation circuit 1. Image signal D from image reading device
It is separated into si and the image signal Dci from the computer.

The image signal Dsi from the image reading device is supplied in parallel to the edge processing filters 2 and 3 and the area identification section 5.

The edge processing filter 2 is a two-dimensional filter for a character image that enhances the Dsi edge component of the image signal, assuming that the target pixel area of the input image is a line part or a character part. The edge processing filter 3 assumes that the pixel region of interest of the input image is a halftone portion and outputs the image signal Ds
This is a two-dimensional filter for halftone images that removes the halftone dot component of i and enhances the band of the image signal Dsi.

The outputs of the edge processing filters 2 and 3 are
The signal is supplied to the selection circuit 4 and switched by the identification signal Se from the area identification section 5 as described later, and is supplied to the subsequent stage as an image signal Dso from the image reading device subjected to the edge processing.

The area identifying section 5 identifies whether the pixel of interest of the input image is a line portion or a character portion or a halftone portion based on the edge degree of the pixel of interest of the input image.

That is, in the area discriminating section 5, in the edge degree detecting circuit 6, the maximum value and the minimum value of the image signal Dsi for each pixel in the area of, for example, 5 × 5 pixels centering on the target pixel of the input image. The difference between the values is detected as the edge degree for the target pixel.

The edge degree is converted into a standardized edge degree signal De having a predetermined number of bits.
Is supplied to the comparison circuit 7 and compared with the reference value signal Dt read from the ROM 8. When the edge degree signal De from the comparison circuit 7 is larger than the reference value signal Dt, it becomes "1". An identification signal Se that becomes “0” is obtained.

When the identification signal Se is supplied as a switching signal to the selection circuit 4 and the identification signal Se becomes "1", the edge processing filter 2 outputs the edge-processed image signal Dso from the image reading device. When the image signal is extracted and the identification signal Se becomes “0”, the image signal from the edge processing filter 3 is extracted as the image signal Dso from the edge-processed image reading device.

[0015] Therefore, in a line portion or a character portion having a high edge degree of an image read by the image reading apparatus, the edge is sharpened and the sharpness is enhanced. The edge of the image signal Dsi from the image reading device is processed so that the image is smoothed and the halftone is reproduced well.

The image signal Dci from the computer is separately supplied to an edge processing unit 9 where edge processing such as smoothing lines and characters by font development is performed as necessary, and supplied to a subsequent stage from the edge processing unit 9. An image signal Dco to be output is output.

[0017]

However, FIG.
In the conventional image processing apparatus described above, the input multi-valued image signal Di is separated into the image signal Dsi from the image reading device and the image signal Dci from the computer by the image separation circuit 1, and the separated image signal Dsi And Dci are subjected to completely different edge processing, so that the image read by the image reading device and the image generated by the computer in the input image for one page represented by the input multi-valued image signal Di When mixed, there is a disadvantage that an unnatural image defect occurs in the output image at the boundary between the two.

In order to avoid such image defects, an input multi-valued image signal is subjected to edge processing by a common edge processing unit without being separated into an image signal from an image reading device and an image signal from a computer. Is also conceivable.
However, for example, for the same character portion, the character portion of the image read by the image reading device needs to have enhanced edges, whereas the character portion of the computer-generated image obtained by font expansion has smooth edges. Therefore, different types of edge processing are required for the image signal from the image reading device and the image signal from the computer.

Therefore, when an input multi-valued image signal is not separated into an image signal from an image reading device and an image signal from a computer but is subjected to edge processing by a common edge processing unit, the image signal from the image reading device and the computer There is a disadvantage that any one or both of the image signals from the camera and the camera must be subjected to inappropriate edge processing.

Furthermore, the conventional image processing apparatus described above
The area identification unit 5 identifies a line portion or a character portion of the image read by the image reading device and a halftone portion from the edge degree of the target pixel. If it is made to be distinguished from a line part or a character part, even a halftone part composed of a halftone dot having a relatively large edge degree will be recognized as a line part or a character part, and the halftone part will be smooth. If it is impossible to reproduce the halftone part, on the contrary, even if the edge degree is relatively large, it is surely identified as the halftone part. As a result, there is a disadvantage that the sharpness of the line portion or the character portion is deteriorated.

JP-A-60-32475 and JP-A-6-32475
Various methods for reliably identifying a line portion, a character portion, or a halftone portion have been proposed in Japanese Patent Application Laid-Open No. 0-103872 and the like, but none of these methods can completely identify them.

In the conventional image processing apparatus, the output of the edge processing filter is switched according to the identification signal Se for the line portion or the character portion and the halftone portion, so that the line portion or the image read by the image reading device is switched. There is a disadvantage that an unnatural image defect occurs in an output image even in a boundary line between a character portion and a halftone portion or in an erroneous determination portion.

Therefore, according to the present invention, an input multi-valued image signal in which an image signal from an image reading device and an image signal from a computer are mixed, as in a complex system having a copying function and a printer function, is edged. In an image processing apparatus for processing, appropriate edge processing is performed on any image or image area without causing unnatural image defects in an output image.

[0024]

According to the first aspect of the present invention, reference numerals in the embodiment shown in FIG.
From the input multi-valued image signal Di, a binary value representing the degree to which the input image has the property of a binary image, which is an image composed of multi-valued image signals having values within a predetermined range near the maximum value or the minimum value, as a continuous amount Property value calculating means 40 for calculating the property value
From the input multi-valued image signal Di, the degree of linear property representing the degree to which the input image has the properties of a character image or a line image as a continuous amount. A non-part portion and a portion where the degree of connection of pixels having similar pixel values is low;
A plurality of edge processing units 21 to 23, each of which performs edge processing on the input multi-level image signal Di and has different frequency characteristics from each other
The plurality of edge processing means 21 to 21 according to the binary property degree calculated by the binary property degree calculation means 40 and the linear property degree calculated by the linear property degree calculation means 30. And a weighted averaging means 50 for averaging the outputs D1 to D3 of the 23 outputs.

According to the second aspect of the present invention, when the reference numerals of the embodiment shown in FIG. 2 and described later are made to correspond, the input multi-level image signal Di indicates that the input image is within a predetermined range near the maximum value or the minimum value. A binary property degree calculating means 40 for calculating a binary property degree representing the degree of having the property of a binary image which is a multi-valued image signal of a value as a continuous amount; and an input multi-valued image signal Di. Is a linear property degree representing the degree to which the input image has the properties of a character image or a line image as a continuous quantity,
In a character portion or a line portion, the degree increases, and in a portion that is not an edge portion and in a portion where the degree of connection of pixels having similar pixel values is low, the degree of linear property decreases to a degree that is small. Means 30, a plurality of edge processing means 21 to 23, each of which performs edge processing on the input multi-valued image signal Di, having different frequency characteristics, and a binary property degree calculated by the binary property degree calculation means 40; Frequency characteristic modulating means 52, 81 to continuously change the frequency characteristics f1 to f3 of the plurality of edge processing means 21 to 23 in accordance with the linear property degree calculated by the linear property degree calculating means 30; 83 and a synthesizing unit 90 for synthesizing the outputs D1 to D3 of the plurality of edge processing units 21 to 23.
And are provided.

[0026]

According to a third aspect of the present invention, in the image processing apparatus of the first or second aspect , the weighted averaging means 50
Alternatively, the frequency characteristic modulating means 52, 81 to 83 have a high binary property degree calculated by the binary property degree calculating means 40 and a linear property degree calculated by the linear property degree calculating means 30. Is higher, the weighting of the outputs D1 to D3 of the plurality of edge processing means 21 to 23 or the frequency characteristics of the plurality of edge processing means 21 to 23 as a whole is set so as to increase the degree of smoothing. .

According to a fourth aspect of the present invention, in the image processing apparatus of the first or second aspect , the weighted averaging means 50
Alternatively, the frequency characteristic modulating means 52, 81 to 83 have a low binary property degree calculated by the binary property degree calculating means 40 and have a linear property degree calculated by the linear property degree calculating means 30. Is higher, the weighting of the outputs D1 to D3 of the plurality of edge processing means 21 to 23 or the frequency characteristics of the plurality of edge processing means 21 to 23 as a whole is set so as to increase the image sharpness. .

According to a fifth aspect of the present invention, in the image processing apparatus of the first or second aspect , the weighted averaging means 50
Alternatively, the frequency characteristic modulating means 52, 81 to 83 have a low binary property degree calculated by the binary property degree calculating means 40 and have a linear property degree calculated by the linear property degree calculating means 30. The lower the value is, the more weighting is applied to the outputs D1 to D3 of the plurality of edge processing units 21 to 23, or the frequency characteristics of the entire plurality of edge processing units 21 to 23 are set so as to increase the degree of smoothing. .

In the present invention, "linear" means
It includes "characteristic".

[0031]

In the image processing apparatus according to the present invention, the outputs D1 to D3 of the plurality of edge processing units 21 to 23, which perform edge processing on the input multi-level image signal Di and have different frequency characteristics, are provided. , Weighted coefficients k1 to k3 corresponding thereto are weighted and averaged in a state where the weighting coefficients k1 to k3 are continuously changed according to the binary property level and the linear property level of the input image, or the frequency characteristics f1 of the plurality of edge processing units 21 to 23 are used. The outputs D1 to D3 of the plurality of edge processing units 21 to 23 are combined in a state where .about.f3 are continuously changed according to the binary property level and the linear property level of the input image.

Therefore, when a binary image such as an image generated by a computer and a non-binary image such as an image read by an image reading apparatus are mixed in an input image, a boundary between the two is obtained. No unnatural image defects occur in the output image in the part, and the boundary or erroneous determination part between the line part or the character part having the linear characteristic of each image and the halftone part having the non-linear characteristic of each image. No unnatural image defects occur in the output image.

[0033]

FIG. 1 shows an example of an image processing apparatus according to the present invention. An input terminal 10 receives a binary image signal from a computer or the like and a non-binary image signal from an image reader or the like as an input multi-valued image signal Di. Eight-bit digital data, in which a value image signal can be mixed, is supplied.

The input multi-valued image signal Di from the input terminal 10 is supplied in parallel to the edge processing filters 21 to 23, the linear property degree calculator 30 and the binary property degree calculator 40.

In this example, the degree of linear property calculation unit 30 determines the degree to which the pixel of interest of the input image has the linear property as a continuous amount based on the edge degree and the connectivity described later for the pixel of interest of the input image. Is calculated. Therefore, the input multi-level image signal Di is supplied in parallel to the edge degree detection circuit 31 and the connectivity degree detection circuit 32 in the linear property degree calculation unit 30.

The edge degree detection circuit 31 outputs an input multi-valued image signal D for each pixel in a 5 × 5 pixel area Wp centered on the target pixel Pc as shown in FIG. 3, for example.
This constitutes a two-dimensional differential filter by a convolution operation of i and a coefficient having a value as shown in FIG. 4 corresponding to FIG. 3, and has a spatial frequency characteristic as shown in FIG. .

The output of the edge degree detection circuit 31 is a continuous value which becomes a large value in a line portion or a character portion of the input image, and becomes a small value in a halftone portion composed of halftone dots. Is normalized to 8-bit data that can take a value from 0 to 255 as the edge degree for the pixel of interest Pc, and is supplied to the synthesizing circuit 33 as the edge degree signal De.

In the connectivity detection circuit 32, the center of the input image is set to a pixel Pc of interest as shown in FIG.
As shown in FIG. 6, the pixel of interest P in the area Wp of × 5 pixels
The values A, B, C, and C of the input multi-level image signal Di for a total of 17 pixels arranged in the horizontal direction, the vertical direction, the oblique direction from the upper left to the lower right, and the oblique direction from the upper right to the lower left centering on c. D, E, F, G, H, I, J,
K. From L, M, N, O, P, and Q, a value represented by r = 255-min (V1, V2, V3, V4) (1) as the connectivity r for the pixel of interest Pc of the input image. Is calculated.

Where the values V1, V2, V
V1 = max (| A−C |, | B−C |, | D−C |, | E−C |)... V2 = max (| F−C |, | G−C) |, | H-C |, | I-C |) ... V3 = max (| J-C |, | K-C |, | L-C |, | M-C |) ... V4 = max (| N −C |, | OC |, | PC |, | QC |).

That is, the value V1 is the value of the target pixel Pc of the five pixels arranged in the horizontal direction around the target pixel Pc.
Is the largest of the absolute values of the differences between the pixel values A, B, D, and E of the pixels except for the pixel value C of the target pixel Pc, and the five absolute values are arranged in the horizontal direction around the target pixel Pc. It shows the magnitude of the change in pixel value between pixels.

Similarly, the value V2 is arranged in the vertical direction around the pixel of interest Pc, the value V3 is arranged in an oblique direction from the upper left to the lower right around the pixel of interest Pc, and the value V4 is arranged in the upper right around the pixel of interest Pc. It shows the magnitude of the change in the pixel value among the five pixels, which are arranged in an oblique direction from to the lower left.

The four maximum values V1 to V4
The connectivity r is the value obtained by subtracting the minimum value from 255 from.

As specific numerical examples, the above pixel values A to Q
Are the values shown in FIG. 7 and the case shown in FIG.

FIG. 7 shows a straight line portion along the horizontal direction. In this case, V1 = max (| 72−71 |, | 74−71 |, | 72−71 |, | 74−71 |) = 3 V2 = max (| 12−71 |, | 26−71 |, | 65−71 |, | 23−71 |) = 59 V3 = max (| 13−71 |, | 26−71 |, | 60−71 |, | 21−71 |) = 58 V4 = max (| 12−71 |, | 26−71 |, | 67−71 |, | 29−71 |) = 59, so that r = 255−min (3,59) , 58, 59) = 255
−3 = 252, and the connectivity r is calculated as a considerably large value.

FIG. 8 shows a halftone portion composed of halftone dots. In this case, V1 = 68, V2 = 64, V3 = 70, and V4 = 28, so that r = 255-28 = 227. The connectivity r is calculated as a smaller value than in the case of FIG.

In the connectivity detection circuit 32, the connectivity r of the target pixel Pc of the input image, which can take a value from 0 to 255 calculated in this way, is synthesized as a connectivity signal Dr of 8-bit data. The signal is output to the circuit 33.

However, a value represented by r = 1 / min (V1, V2, V3, V4) (2) is calculated as the connectivity r instead of the equation (1), and the calculated value is 0 to 255
The data may be normalized to 8-bit data that can take values up to and supplied to the combining circuit 33 as the connectivity signal Dr.

By the way, in this case, the difference in the connectivity r between the line portion or the character portion and the halftone portion becomes larger than when the expression (1) is used. .

In any case, the output of the connectivity detection circuit 32 is also obtained as a continuous amount having a large value in the line portion or the character portion of the input image and a small value in the halftone portion.

In the synthesizing circuit 33, the edge degree signal De from the edge degree detecting circuit 31 and the connectivity detecting circuit 3
2, the value of the output composite signal changes in accordance with the value of the connectivity signal Dr from step 2, and the greater the value of both the edge degree signal De and the connectivity signal Dr, the greater the value of the output composite signal. And the output composite signal is from 0 to 25
It is obtained by being normalized to 8-bit data that can take values up to 5.

Therefore, the output of the synthesizing circuit 33 is also obtained as a continuous amount having a large value in the line portion or the character portion of the input image and a small value in the halftone portion.

The output of the synthesizing circuit 33 is
4 and is averaged for each pixel of the input image in a region of, for example, 5 × 5 pixels centered on the pixel. This brings about an effect of diffusing a region having a high degree of linear property to a peripheral region.

The 8-bit data output of the averaging circuit 34 finally determines the linear property s for each pixel of the input image.
Is continuously supplied to the weighting coefficient calculation circuit 52 of the weighted averaging unit 50.

In this example, the binary property degree calculating section 40 calculates the target pixel Pc of the input image as shown in FIG.
From the value of the input multi-level image signal Di for each pixel in the 5 × 5 pixel area Wp centered at, for example, the pixel value becomes 0 or near 255, for example, from 0 to 9, or 24.
The number of pixels N1 in the region Wp having a value from 6 to 255
Of the number of pixels N0 (= 25) in the area Wp
1 / N0 is calculated as a continuous amount from 0 to 1 as the binary property degree n of the target pixel Pc of the input image.

A character image or the like generated by font expansion inside a computer has a binary property, and has a region Wp
As shown in FIG. 9, the pixel values in the area are concentrated around 0 or 255, and the ratio N1 / N0 becomes a large value close to 1.

On the other hand, a natural image or the like read by the image reading device has a non-binary property, and the pixel values in the area Wp are dispersed as shown in FIG. /
N0 is a small value close to 0.

In the binary property degree calculating section 40, the ratio N1 / calculated as a continuous amount from 0 to 1 is obtained.
N0, that is, the binary property degree n of the target pixel Pc of the input image is normalized to 8-bit data that can take a value from 0 to 255, and the weighted averaging unit 50 as the binary property degree signal Sn. Is supplied to the weight coefficient calculation circuit 52.

However, since the total number of pixels N0 in the area Wp is predetermined as N0 = 25 in the above example, the above ratio N1 / N0 is not calculated.
The number N1 of pixels in the region Wp having a pixel value of 0 or around 255 is directly calculated as a continuous property from 0 to, for example, 25 as the binary property degree n of the pixel of interest Pc of the input image. The value may be normalized to 8-bit data that can take a value from 0 to 255 and output as the binary property signal Sn.

The edge processing filters 21, 22, and 23
For example, in a 5 × 5 pixel window,
The edge processing is performed on the input multi-level image signal Di by a convolution operation between the input multi-level image signal Di and the filter coefficient read from the memory 60. However,
The filter coefficients for the 5.times.5 pixels are different between the edge processing filters 21, 22, and 23, whereby the frequency characteristics of the edge processing filters 21, 22, and 23 are different from each other.

More specifically, the edge processing filter 21 is for characters, and its frequency characteristic is to enhance high frequency components as shown by a characteristic f1 in FIG. The edge processing filter 22 is for a binary picture, and has a frequency characteristic that allows the input signal to pass almost as it is, as shown by a characteristic f2 in FIG. The edge processing filter 23 is for halftone and binary characters, and its frequency characteristic is the characteristic f3 of FIG.
As shown in FIG. 7, high-frequency components are cut to remove moire and noise.

The edge processing filters 21, 22, 2
The filter coefficient for 3 is generated in the control unit 70 having the CPU 71 in advance, and is written from the control unit 70 to the memory 60.

Image signals D 1 and D made of 8-bit data from the edge processing filters 21, 22 and 23, respectively.
2 and D3 are supplied to the product-sum operation circuit 51 of the weighted averaging unit 50.

In the weighted averaging unit 50, in the weight coefficient calculation circuit 52, the binary property degree signal Sn from the binary property degree calculation unit 40 and the linear property degree from the linear property degree calculation unit 30 From signal Ss, edge processing filter 2
Weight coefficients k1, k2, and k3, which will be described later, with respect to the image signals D1, D2, and D3 from 1, 22, and 23 are calculated, and in the product-sum operation circuit 51, Do = k1 · D1 + k2 · D2 + k3 · D3 (3) The edge processing filters 21 and
The image signals D1, D2 and D3 from the circuits 22 and 23 are weighted and averaged, and an output multi-valued image signal Do composed of 8-bit data is obtained from the product-sum operation circuit 51.

The local area of the input image is roughly divided into
(A) a portion having a low binary property level and a high linear property level, such as a line portion or a character portion of an image read by the image reading device, and (b) an image read by the image reading device. A portion having a low binary property and a low linear property, such as a halftone portion composed of halftone dots; and (c) a portion having a low linear property, such as a line portion or a character portion of an image generated by a computer. (D) a portion having a high binary property and a high linearity, such as a halftone portion composed of halftone dots of an image generated by a computer; The degree of property is divided into four types, low.

In the weighting coefficient calculation circuit 52, the value of the binary property degree signal Sn is used as the weighting coefficient k1 for the image signal D1 from the edge processing filter 21 according to the difference in the properties of the local area of the input image. Is smaller and the value of the linear property signal Ss is larger, that is, as shown in FIG. 12, the lower the binary property of the target pixel of the input image and the higher the linear property, Values that increase as the value approaches 1 are calculated.

Image signal D from edge processing filter 22
2, the value of the binary property signal Sn is large and the linear property signal S
As the value of s is smaller, that is, as shown in FIG. 13, the binary property degree of the target pixel of the input image is higher and the linear property degree is lower, and the value approaches 1 and becomes larger. Is calculated.

Image signal D from edge processing filter 23
The weight coefficient k3 for 3 is calculated by an operation represented by the following equation: k3 = 1− (k1 + k2) (4) Therefore, the value of the binary property degree signal Sn is large as the weight coefficient k3,
As the value of the linear property signal Ss is larger, or the value of the binary property signal Sn is smaller and the value of the linear property signal Ss is smaller, that is, as shown in FIG. The value approaches 1 as the binary property level of the pixel of interest of the input image is higher and the linear property level is higher, or as the binary property level is lower and the linear property level is lower. Is calculated.

The characteristics of the weighting factors k1 to k3 with respect to the binary property level signal Sn and the linear property level signal Ss can be set or adjusted by a control signal CT from the control unit 70.

Therefore, the ratio of the image signal D1 from the edge processing filter 21 to the output multivalued image signal Do in the line portion or the character portion of the image read by the image reading device in the above (a) becomes large, That is, equivalently, the frequency characteristics of the entire edge processing filters 21 to 23 become close to the characteristics f1 shown in FIG. 11, and the edges of the line portion or the character portion of the image read by the image reading device are emphasized, and the sharpness is sharpened. Is enhanced.

In the halftone section composed of the halftone dots of the image read by the image reading device in the above (b), the ratio of the image signal D3 from the edge processing filter 23 to the output multi-valued image signal Do is large. That is, equivalently, the frequency characteristics of the entire edge processing filters 21 to 23 become close to the characteristic f3 shown in FIG. 11, and the halftone portion composed of the halftone dots of the image read by the image reading device is smooth. And the half tone is reproduced well.

In the line portion or the character portion of the image generated by the computer in (c), similarly to the halftone portion formed by the halftone dot of the image read by the image reading device in (b), The ratio of the image signal D3 from the edge processing filter 23 to the output multi-valued image signal Do increases,
That is, equivalently, the frequency characteristics of the entire edge processing filters 21 to 23 become close to the characteristics f3 shown in FIG. 11, and unnatural steps or jags in line portions or character portions of an image generated by the computer are reduced. , Lines or characters are output smoothly.

In the halftone portion composed of the halftone dots of the image generated by the computer in the above (d), the ratio of the image signal D2 from the edge processing filter 22 to the output multi-valued image signal Do becomes large. That is, equivalently, the frequency characteristics of the entire edge processing filters 21 to 23 become close to the characteristic f2 shown in FIG. 11, and the halftone portion composed of the halftone dots of the image generated by the computer is output almost as it is. .

As described above, according to the example shown in FIG. 1, appropriate edge processing is performed on any image or image area in accordance with the characteristics of each image or image area, and the edge processing filters 21 to 21 are used. 23 from the image signal D
Since the weighting coefficients k1 to k3 for 1 to D3 are continuously changed according to the binary property level and the linear property level of the input multi-valued image signal Di, the image read by the image reading device and the image generated by the computer are generated. An unnatural image defect does not occur in the output image at the boundary between the output images and at the boundary between the line portion or the character portion of each image and the halftone portion composed of halftone dots.

The binary property degree calculating section 40 calculates the binary property degree of the pixel of interest from the frequency distribution of pixel values in a predetermined number of pixels around the pixel of interest in the input image. Therefore, the binary property degree is accurately and easily calculated, and the linear property degree calculation unit 30 calculates the linear property degree of the target pixel from the edge degree and the connectivity of the target pixel of the input image. Is calculated, the linear property degree is accurately and stably calculated.

FIG. 2 shows another example of the image processing apparatus according to the present invention. The filter coefficients read from the memory 60 for the edge processing filters 21, 22, and 23 are modulated by modulation circuits 81, 82, and 83, respectively. , The same weighting factors k1, k1 as in the example of FIG.
modulated by k2 and k3, and the edge processing filter 2
1, 22, 23, and the edge processing filters 21,
The image signals D1, D2, and D3 from the circuits 22 and 23 are added with the same weight in the adding circuit 90, so that the adding circuit 90 obtains the output multi-valued image signal Do as 8-bit data. The modulation circuits 81 to 83 are each configured by a multiplier.

The other structure is the same as that of the example shown in FIG.
The linear property degree calculating section 30 is also configured in the same manner as in the example of FIG.

In this example, the weighting factors k1, k2,
Filter coefficients for the edge processing filters 21, 22, and 23 are modulated in accordance with k3, thereby modulating the frequency characteristics of the edge processing filters 21, 22, and 23, and changing the frequency characteristics of the entire edge processing filters 21 to 23. . Therefore, according to this example, the same effect as the example of FIG. 1 can be obtained.

Although common to the examples of FIGS. 1 and 2,
The pixel size of the convolution operation in the edge processing filters 21 to 23 is not limited to 5 × 5 pixels, but is 3 × 3.
Pixels, 3 × 5 pixels, 5 × 7 pixels, 7 × 7 pixels, or the like may be used.

The pixel size of the convolution operation in the edge degree detection circuit 31 of the linear property degree calculation section 30 is also
Similarly, it is not necessary to limit to 5 × 5 pixels.

The connectivity detection circuit 3 of the linear property calculation unit 30
In FIG. 2, one side of each of the horizontal direction, the vertical direction, the oblique direction from the upper left to the lower right, and the oblique direction from the upper right to the lower left centered on the target pixel Pc as shown in FIG. V1 = max (| AC |, | BC |) V2 = max (| DC |, | EC |) V3 = max (| FC |, | GC |) V4 = max (| H−C |, | I−C |) V5 = max (| J−C |, | K−C |) V6 = max (| L−C |, | M−C |) V7 = max (│NC│, │OC│) V8 = max (│PC│, │QC│) is calculated, and as the connectivity r for the target pixel Pc, r = 255-min ( V1, V2, V3, V4, V5
V6, V7, V8) or r = 255 / min (V1, V2, V3, V4, V5
V6, V7, V8) may be calculated.

The pixel size for which the connectivity is to be calculated in the connectivity detection circuit 32 need not be limited to 5 × 5 pixels.

Furthermore, the output Dr of the connectivity detection circuit 32 may be averaged in a region of a fixed pixel size, and the averaged connectivity signal may be supplied to the synthesis circuit 33.

Although the examples of FIGS. 1 and 2 are provided with three edge processing filters 21 to 23, it is sufficient that the number of edge processing filters is two or more. When only two edge processing filters are provided, one of them is used for characters as in the above-described edge processing filter 21, and its frequency characteristic is enhanced as shown by a characteristic f1 in FIG. The other one is used for halftone and binary characters as in the above-described edge processing filter 23, and its frequency characteristic is to cut high frequency components as shown by the characteristic f3 in FIG. The frequency characteristic that allows the input signal to pass almost as it is, as in the above-described edge processing filter 22, may be obtained as a composite frequency characteristic of one and the other edge processing filters.

The present invention can be applied to a color image signal. In the case of a color image signal, it is not necessary to perform the above-described edge processing on the multi-valued image signal of each color component. For example, in the case of a color image signal expressed in the Lab color space, the above-described processing is performed only on the L signal which is a brightness signal. The edge processing is performed, and the chromaticity signals a and b may be delayed so as to match the time with the edge-processed L signal.

[0085]

As described above, according to the present invention, both the image read by the image reading apparatus and the image generated by the computer can be appropriately adjusted in accordance with the properties of each image or image area. Edge processing and the boundary between the image read by the image reading device and the image generated by the computer, and the boundary between the line or character part of each image and the halftone part composed of halftone dots. No unnatural image defect occurs in the output image in the section.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an example of an image processing apparatus according to the present invention.

FIG. 2 is a block diagram showing another example of the image processing apparatus of the present invention.

FIG. 3 is a diagram provided for describing edge degree detection in a linear property degree calculation unit in each example.

FIG. 4 is a diagram provided for describing edge degree detection in a linear property degree calculation unit in each example.

FIG. 5 is a diagram illustrating an example of a spatial frequency characteristic of an edge degree detection circuit of a linear property degree calculation unit in each example.

FIG. 6 is a diagram provided for explanation of connection degree detection in a linear property degree detection unit in each example.

FIG. 7 is a diagram provided for describing connectivity detection in a linear property detection unit in each example.

FIG. 8 is a diagram provided for explanation of connectivity detection in a linear property detection unit in each example.

FIG. 9 is a diagram provided to explain a binary property degree calculation in a binary property degree calculation unit in each example.

FIG. 10 is a diagram provided for explaining a binary property degree calculation in a binary property degree calculation unit in each example.

FIG. 11 is a diagram illustrating an example of frequency characteristics of three edge processing filters of each example.

FIG. 12 is a diagram illustrating an example of a change characteristic of a first weighting coefficient in each example.

FIG. 13 is a diagram illustrating an example of a change characteristic of a second weighting coefficient in each example.

FIG. 14 is a diagram illustrating an example of a change characteristic of a third weight coefficient in each example.

FIG. 15 is a block diagram illustrating an example of a conventional image processing apparatus.

[Explanation of symbols]

 21 to 23 edge processing filter (edge processing means) 30 linear property degree calculating section (linear property degree calculating means) 40 binary property degree calculating section (binary property degree calculating means) 50 weighted averaging section (weighted) Averaging means) 52 Weight coefficient calculation circuit (frequency characteristic modulation means) 81-83 Modulation circuit (frequency characteristic modulation means) 90 Addition circuit (synthesis means)

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 1/40-1/409 G06T 1/00-7/00

Claims (5)

(57) [Claims] 1. A method for determining, from an input multi-valued image signal, a degree to which an input image has the property of a binary image which is an image composed of multi-valued image signals having values within a predetermined range near a maximum value or a minimum value. A binary property degree calculating means for calculating a binary property degree represented as: a linear property degree representing a degree of a character image or a line image having a property of a character image or a line image as a continuous quantity from an input multi-valued image signal. In a character part or a line part, the degree becomes large, and in a part which is not an edge part and a part where pixels having similar pixel values are low in degree of connection, the degree becomes small. Degree calculating means, a plurality of edge processing means, each of which performs edge processing on the input multi-valued image signal, having different frequency characteristics from each other; the binary property degree calculated by the binary property degree calculating means; Property degree calculation According to the linear nature degree calculated by the step, the image processing device characterized by and a weighted averaging means for weighting averaging the outputs of said plurality of edge processing means. 2. A method according to claim 1, wherein the input image has a continuous value representing the degree to which the input image has the property of a binary image which is an image composed of multi-valued image signals having values within a predetermined range near a maximum value or a minimum value. A binary property degree calculating means for calculating a binary property degree expressed as: a linear property degree representing a degree of a character image or a line image having a property of a character image or a line image as a continuous quantity from an input multi-valued image signal. In a character portion or a line portion, the degree is large, and in a portion that is not an edge portion and a portion where pixels with similar pixel values are low in degree of connection, the degree is reduced. Degree calculating means, a plurality of edge processing means, each of which performs edge processing on the input multi-valued image signal, having different frequency characteristics from each other; the binary property degree calculated by the binary property degree calculating means; Property degree calculation Frequency characteristic modulating means for continuously changing the frequency characteristics of the plurality of edge processing means in accordance with the degree of linear property calculated by the step; and synthesizing means for synthesizing outputs of the plurality of edge processing means. An image processing apparatus comprising: 3. The image processing apparatus according to claim 1, wherein the weighted averaging means or the frequency characteristic modulating means has a high binary property degree calculated by the binary property degree calculating means, And, the higher the linear property degree calculated by the linear property degree calculating means, the higher the weighting of the outputs of the plurality of edge processing means or the higher the degree of smoothing of the frequency characteristics of the plurality of edge processing means as a whole. An image processing apparatus, wherein the direction is set. 4. The image processing apparatus according to claim 1, wherein the weighted averaging means or the frequency characteristic modulating means has a low binary property degree calculated by the binary property degree calculating means, And, the higher the linear property degree calculated by the linear property degree calculation means, the higher the weight of the outputs of the plurality of edge processing means or the frequency characteristics of the plurality of edge processing means as a whole. An image processing apparatus, wherein the direction is set. 5. The image processing apparatus according to claim 1, wherein the weighted averaging means or the frequency characteristic modulation means has a low binary property degree calculated by the binary property degree calculation means, In addition, the lower the linear property degree calculated by the linear property degree calculation means, the higher the degree of smoothing the weighting of the outputs of the plurality of edge processing means or the frequency characteristics of the plurality of edge processing means as a whole. An image processing apparatus, wherein the direction is set.