KR100998374B1 - Image Processing Method and Apparatus for Detecting the Starting Point and End Point of the Lines - Google Patents

Abstract

The present invention relates to an image processing method and apparatus for detecting a start point and an end point of a straight line. The present invention relates to a method for generating and storing a contour image using external image data input from the outside and converting the contour image by half-converting the generated contour image. Detecting a variable, and detecting a start point and an end point of a straight line by matching the contour image with an infinite straight line image represented by a straight line parameter, thereby providing a size and shape of an actual object. It is possible to accurately determine the characteristics of, thereby realizing the recognition of real objects efficiently.

Half transform, straight line detector, contour image, edge image

Description

Image Processing Method and Apparatus for Detecting the Starting Point and End Point of the Lines}

The present invention relates to an image processing method and apparatus for detecting a starting point and an ending point of a straight line.

Computer vision, which analyzes an image and recognizes objects or environments in the image, has various applications. In performing recognition of an object or environment, important information is straight line information. The image processing apparatuses may recognize an object based on a straight line and detect a lane on the road.

The half transform is used as a method for detecting a straight line existing in an image. The half transform transforms an original image into an edge image having contour information, and then extracts a parameter of straight lines that all edge points of the transformed edge image affect. The linear detection through the half transform is widely used when there is noise in an image or robust against a change in brightness.

However, only the parameter detection of a straight line obtained through the half transform is difficult to recognize an object by using information of the actual straight line or apply it directly to an application field such as a robot or a vehicle. This is because the parameters obtained through the half transform do not know at which point the line actually starts and ends at which point.

Accordingly, the present invention is to solve the problems according to the prior art, an image processing method for detecting the exact starting point and end point of the straight line using the contour image obtained from the original image and the straight line parameters that are the result of the half transform and The provision of the device is for that purpose.

An image processing method for detecting a start point and an end point of a straight line according to an aspect of the present invention generates and stores an edge image using external image data input from the outside, and half-converts the generated contour image. Thereby detecting a straight line parameter and detecting a start point and an end point of an infinite straight line image represented by the detected straight line parameter by using the stored contour image.

The storing of the contour image may include storing coordinates of valid pixels constituting the generated contour image.

The detecting of the starting point and the ending point of the infinite linear image may include matching the infinite linear image represented by the detected linear parameter with the stored contour image, and matching the contour image among pixels constituting the infinite linear image. The method may include acquiring only pixels and checking start and end points of a line segment composed of matching pixels and acquiring pixel information thereof.

In this case, a process of matching an infinite straight line and a contour image to obtain a pixel matching the contour image among the infinite straight line images, and checking a start point and an end point of a line segment composed of the pixels is performed for each linear parameter.

The detecting of the straight line parameter may include detecting the straight line parameter by using a result of repeatedly calculating the following equation for each of a plurality of edge points for each value having a predetermined range.

Here, x and y are the x and y coordinate values of each edge point.

In this case, the step of repeatedly calculating the equation for each value having a predetermined range may be performed in parallel at a plurality.

The detecting of the linear parameter may include increasing a value of a parameter cell corresponding to a ρ, θ pair obtained from an equation by one and if the value of each cell of the parameter memory is greater than or equal to a predetermined value, And determining a parameter corresponding to the variable memory cell as a valid linear parameter. In this case, the detecting of the straight line parameter may include searching for a cell having the largest value among a plurality of parameter memory cells corresponding to a certain area of the parameter space, and validating the cell if the found cell value is larger than a reference value. The method may further include determining a parameter.

According to another aspect of the present invention, an image processing apparatus for detecting a start point and an end point of a straight line may include: an edge generator configured to generate an edge image using external image data input from the outside and store the edge image in a memory; Starting point of the infinite linear image represented by the detected linear parameter using the half transform unit for obtaining the linear parameters by performing the half transform of the pixel coordinate information of the contour image generated by the generator and the contour image stored in the memory unit And end point.

The matching unit acquires only the pixel matching the contour image among the pixels constituting the infinite linear image by matching the infinite linear image represented by the linear parameter with the stored contour image, and starts the line of the segment consisting of the matching pixels. And check the end point.

The matching unit may acquire a pixel matching the contour image of the infinite linear image by matching the infinite straight line and the contour image, and check the start point and the end point of the line segment composed of the pixel for each linear parameter. .

In addition, the image processing apparatus according to the present invention includes an effective parameter detector for detecting valid linear parameters among the half-converted linear parameters, and a contour image for storing coordinates of valid pixels constituting the contour image generated by the edge generator. The apparatus may further include a valid linear parameter list for storing the valid linear parameters checked by the storage unit and the valid parameter detection unit.

The half transform unit performs a half transform by calculating the following equation for each value having a predetermined range with respect to a plurality of edge points.

Here, x and y are the x and y coordinate values of each edge point.

In addition, it is more preferable that the half transform unit performs a plurality of operations of the equation in parallel at one time. To this end, the half transform unit may include a plurality of trigonometric and arithmetic operators.

The half transform unit increases the value of the parameter cell corresponding to ρ and θ obtained from the equation by one. Accordingly, when the value of each cell of the parameter memory is greater than or equal to a predetermined value, the valid parameter detector determines a parameter corresponding to the parameter memory cell as a valid linear parameter. In addition, the valid parameter detector may search for a cell having the largest value among a plurality of parameter memory cells corresponding to a predetermined area of a parameter space, and if the searched cell value is larger than a reference value, the effective linear parameter You can also judge.

According to the image processing method and apparatus for detecting the start point and end point of a straight line according to the present invention as described above, it is possible to determine the start and end of the image straight line, and has only the straight line information in the field such as object recognition Also, the size and shape of the real object can be determined. The present invention may be applied to an intelligent robot that drives based on image information photographed through a camera and a vehicle driving assistance device that recognizes a lane.

Hereinafter, an image processing method and apparatus for detecting a start point and an end point of a straight line according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating a configuration of an image processing apparatus according to an embodiment of the present invention.

As illustrated in FIG. 1, the image processing apparatus 100 connected to the camera 200 operates as an image inputter 110, an edge generator 120, a half transformer 130, and an effective parameter detector 140. , A matching unit 150, an output unit 160, a memory unit 170, and the like.

The camera 200 refers to a general apparatus for taking an external image by receiving light through a lens or the like. The camera 200 of FIG. 1 transmits the captured external image to the image input unit 110 of the image processing apparatus 100 according to the present invention through a predetermined wired or wireless communication method.

The image input unit 110 receives external image information transmitted from the camera 200, converts the image into digital data, and generates a synchronization signal using its own signal or an output signal of the camera 200. .

The edge generator 120 generates an edge image using external image data in digital form transmitted from the image input unit 110, and calculates coordinates of valid pixels constituting the generated contour image. The method of generating the contour image will not be described in detail.

The contour image generated as described above is stored in the memory unit 170. The area of the memory unit 170 in which the contour image is stored will be referred to as the contour image storage unit 171.

The half transform unit 130 performs half transform with the coordinate information of the contour pixel calculated by the edge generator 120. The result according to the half conversion result is stored in the parameter cell 172 of the memory unit 170.

The effective parameter detector 140 detects a straight line of an image by checking a valid pair of parameters based on a half result stored in the parameter cell 172. The detected straight line corresponds to an infinite straight line having no starting point and no ending point. In addition, the detected valid linear parameter pair may be stored in the valid parameter list 173 of the memory unit 170.

The matching unit 150 corresponds to a component that detects the starting point and the ending point of the straight line by matching the effective linear parameter with the contour image. As such, when the starting point and the ending point of the straight line are detected, information on a line segment that is a finite straight line can be obtained.

The output unit 160 may output a drawn image or the like based on the information about the finite straight line segment obtained by the matching unit 150 or the finite straight line segment information according to the user's request.

2 is a diagram illustrating an image processing method according to another exemplary embodiment of the present invention.

The image processing apparatus 100 receives an external image from an external device such as the camera 200 in operation S201.

The image processing apparatus 100 converts the input external image into digital data, generates a contour image, and stores the information in a memory in order to detect a start point and an end point of an infinite straight line later (S202).

The image processing apparatus 100 performs half transform on the basis of the pixel information of the generated contour image (S203). Thereafter, the image processing apparatus 100 performs a process of obtaining a valid linear parameter based on the result of performing the half transform (S204).

The image processing apparatus 100 forms an infinite linear image through the obtained effective linear parameters, matches the contour image stored in step S202, detects the start point and the end point of the infinite line, and restores the finite linear image. Acquire (S205).

The image processing apparatus 100 outputs, according to a request of the user, the contour image reconstructed based on the acquired finite linear image information or the obtained finite linear image information (S206).

As described with reference to FIG. 2, the image processing apparatus 100 according to the present invention stores information about a valid straight line parameter and a start point and an end point of an infinite straight line restored to the valid straight line parameter, and restores the same image as the contour image later. Can be.

3 is a diagram illustrating an example of a straight line for performing a half transform.

The entire image of FIG. 3 may be expressed through two-dimensional information consisting of the X-axis in the horizontal direction and the Y-axis in the vertical direction, and unique brightness information of pixels corresponding to each point.

It is assumed that the entire image of FIG. 3 has a resolution of 640 × 480 for convenience of description. Of course, the resolution of the entire image is not limited to a specific value, and can be freely changed.

The straight line L of FIG. 3 may be expressed through Equation 1 below represented by an equation of a polar form.

[Equation 1]

Where ρ is the distance between the origin and the straight line of the entire image with coordinates (0,0), and θ is the angle between the normal of the X-axis and the straight line L. Equation 1 can express a straight line of a two-dimensional plane using ρ and θ, the equation using the ρ and θ coordinates is called the polar form. After all, two parameters of (ρ, θ) can represent an infinite straight line existing in the two-dimensional plane.

On the other hand, the range of ρ and θ parameters for expressing the straight line in all cases present in the 640 × 480 image is shown in Equation 2 below. In Equation 2 below, H variable is a vertical resolution value of an image, and W variable is a horizontal resolution value of an image.

The negative value of ρ corresponds to the case where a straight line exists in the third quadrant. Of course, the range of θ may be set to 0 <θ <2π or -π <θ <π to indicate a straight line located in the third quadrant. However, the efficiency of the half transform can be set as shown in Equation 2 below. Express

[Equation 2]

The image processing method according to the present invention is a process of detecting the straight line L from the edge points A1, A2, A3 shown in FIG. In particular, the present invention measures the frequency of occurrence of the edge points A1, A2, A3 to detect a valid straight line. The more linear components, the more edge points located on the straight line appear and the concept of detecting the straight line by measuring the frequency of occurrence of these edge points.

4 is a diagram briefly illustrating a half transform result by the image processing apparatus according to the present invention.

The two-dimensional plane of FIG. 4 is composed of a ρ axis and a θ axis. In particular, the two-dimensional plane of FIG. 4 will be referred to as parameter space. The half transform result of FIG. 4 represents straight lines that can pass through an edge point as a parameter of (ρ, θ).

For example, if the half transform is performed on the edge points A1, A2, and A3 shown in the straight line L of FIG. 3, three sinusoidal curves appear in the parameter space. Here, one sinusoidal curve is a set of (ρ, θ) coordinates of straight lines that can pass through one edge point. In particular, the sinusoidal curve obtained as a result of performing the half transform on the A1 edge point of FIG. 3 means a plurality of straight lines passing through the edge point A1.

The half transform is performed on three edge points A1, A2, and A3 in the two-dimensional plane of FIG. Here, the line represented by the parameter (ρ, θ), that is, the parameter (ρ1, θ1) of the point where the three sinusoidal curves coincide, is a straight line passing through all three edge points (A1, A2, A3). It corresponds to L straight line of 3.

The fast half-conversion linear detection apparatus according to the present invention can detect an infinite straight line passing through an edge point from information of a plurality of edge points using this characteristic.

The image processing apparatus 100 according to the present invention performs a half transform for each edge point to obtain a sinusoidal curve for each edge point. After that, the point where the sinusoidal curve passes is checked and the value of the memory cell corresponding to the point is increased by one. As a result, the memory cell corresponding to one point where the three curves meet (P Point in FIG. 4) has a value of 3, and the memory cell corresponding to the remaining curve part is 1, which corresponds to the point of the region where the curve does not pass. The memory cell will have a value of zero. The image processing apparatus 100 according to the present invention may obtain a straight line passing through an edge point by checking memory cells having a predetermined reference value or more and determining (ρ, θ) to which the memory cell points.

Referring to this operation, the half transform means converting the image space of FIG. 3 into the parameter space of FIG. 4. It is very difficult to know which straight line the edges correspond to in the image space of FIG. 3. When this is converted to the parameter space of FIG. 4 to detect a memory cell having a high value in the parameter space, that is, the parameter (ρ, θ) of the straight line, the straight line formed by the edge point can be clearly known.

5 is a diagram illustrating an example of an operation method of a half transform performed by the image processing apparatus of the present invention.

In the half transform of FIG. 5, the image processing apparatus 100 repeatedly calculates ρ by performing the same operation as [Equation 1] with respect to the coordinates of all edge points by the range of θ. The ρ value thus obtained is matched with the value of θ and reflected in the parameter space of FIG. 4.

The value of the memory cells corresponding to (ρ, θ) according to the result of performing the operation of Equation 1 is increased by 1, and the brightness of the parameter space is changed according to the values of the memory cells.

However, since the half transform of the scheme described with reference to FIG. 5 repeats the operation by the range of θ (No. 315 in FIG. 5), the operation execution time becomes long and the memory usage for storing the parameters also increases. .

6 is a diagram illustrating another example of an operation method of the half transform performed by the image processing apparatus of the present invention.

The gist of the half transform method presented through FIG. 6 is that a plurality of [Equation 1] operations, that is, parallel operations of [Equation 1] are performed at a time. For convenience of description, it is assumed in FIG. 6 that 15 operations are performed at once, that is, in parallel.

Since the image processing apparatus 100 performs 15 Equation 1 operations at a time, the time required for half conversion is reduced by 1/15. In the half transformation method of FIG. 5, a range of θ variables, that is, 315 repetitive operations are performed on one contour image. In contrast, according to the operation of the half transform of FIG. 6, the image processing apparatus 100 only needs to perform 21 repetitive operations.

In conclusion, when the half transform operation method of FIG. 6 is used, the number of repetition operations is shortened, and thus the half transform execution time is shortened. In addition, the memory for storing the value according to the (ρ, θ) coordinates is reduced.

According to the half transform method of the present invention, the space for storing the matching count information according to the (ρ, θ) coordinates is not one memory having a total size of θ × ρ. In the present invention, 15 memories having a size of 1 × ρ are required.

As a result, since the half conversion according to the present invention can be implemented with only a small amount of memory, it is excellent in terms of implementation cost and efficiency. However, the fast half conversion method described in FIG. 6 is difficult to operate in a general PC. Since the PC performs serial processing, 15 operations cannot be performed at the same time. That is, even if the equations described in FIG. 6 are programmed, the PC does not perform 15 operations in parallel but sequentially, and thus the operation execution time cannot be significantly reduced.

7 is a diagram illustrating a hardware configuration of a half converter for performing an operation of a half transform.

The half converter 130 of FIG. 7 finally outputs p when an initial value of [x, y, θ] is input.

As shown in FIG. 7, the half converter 130 includes an input unit 131, a cosine operator 132, a subtractor 133, a sine operator 134, a multiplier 135, an adder 136, and an output unit 137. It can be configured through.

Referring back to FIG. 3, the XY two-dimensional plane assumed in the present invention has a value from 1 to 640 in the horizontal (X coordinate) and 1 to 480 in the vertical (Y coordinate). At this time, in order to reduce the size of the parameter space, the center of the image (320,240) is used as the origin to adjust the horizontal values (-319 to 320) and the vertical values (-239 to 240). Performing this role are the two subtractors 133 shown in FIG.

Meanwhile, the cosine operator 132 and the sine operator 134 respectively receive θ values and output cosθ and sinθ values. The multiplier 135 performs an operation of multiplying x-320 = x 'and a cosθ value and a multiplication of y-240 = y' and a sinθ value. The adder 136 plays a role of adding x'cosθ and y'sinθ values. Through the configuration of the half converter 130 of FIG. 7, the value of ρ may be calculated by performing the equation (1).

The output unit 137 increases the value of the parameter cell 172 corresponding to the input θ and the calculated ρ by one.

Here, the half transform described in FIG. 5 may also be implemented through one half converter 130 of FIG. 7. However, in order to perform the half transform described with reference to FIG. 6, a plurality of half converters 130 of FIG. 7 must be configured in parallel.

8 is a diagram illustrating a structure of a parameter cell in which a half transform result of FIG. 6 is stored.

As mentioned above, in order to store the result of the fast half transform of FIG. 6, the image processing apparatus 100 according to the present invention may include a parameter cell 172 having a structure of 1 × ρ.

The parameter cell 172 of FIG. 8 having a structure of 1 × ρ will be referred to as a split parameter cell in order to distinguish it from the θ × ρ parameter cell required for the half transform of FIG. 5.

Assuming a resolution of 640x480, ρ can have a value from -399 to 400, so a parameter cell 172 with a total of 800 columns is needed. Referring to FIG. 8, each cell having an address from 0 to 799 corresponds to ρ.

When an arbitrary value of p1 is output from the half converter 130 of FIG. 7, the value of the splitting parameter cell according to the value of p1 is increased by one. Specifically, the value of the memory cell according to the value of p1 is stored in the temporary buffer 174. The value of the temporary buffer 174 may be automatically increased by 1 to be written in the original parameter 172.

In this manner, the splitting parameter memory continually performs an operation of incrementing by one with respect to the parameter cell 172 according to the value of ρ which is the result of the calculation of the half converter 130 for each θ.

9 is a diagram illustrating a method of performing a half transform of FIG. 5 by an image processing apparatus according to the present invention.

First, the image processing apparatus 100 according to the present invention initializes values according to its parameter memory (S901). That is, the value of all parameter cells is set to zero.

The image processing apparatus 100 sets the θ value to 1 (S902). Thereafter, a mathematical expression of S903 is performed using the x, y coordinates of the input edge point and the current θ value (S903).

The image processing apparatus 100 increases the value of the parameter memory cell referred to by the current θ value and ρ value by 1 (S904).

After that, check whether the current value of θ is greater than 315 (S905), and if the value of θ is not greater than 315, increase the value of θ by 1 and repeat steps S903 to S905. If the current θ value is greater than 315 in step S905, the half transform is terminated.

10 is a diagram illustrating a method of performing a half transform of FIG. 6 by an image processing apparatus according to the present invention.

First, the image processing apparatus 100 sets the θ value to 1 (S1001). Thereafter, the image processing apparatus 100 according to the present invention initializes values according to its parameter memory (S1002). That is, the value of all parameter cells is set to zero.

After initializing the values of the parameter cells of S1002, the image processing apparatus 100 calculates 15 equations described in FIG. 10 in parallel (S1003).

The image processing apparatus 100 increases the value of the parameter memory cells respectively referred to by 15 pairs of (ρ, θ) for which 15 equations are calculated (S1004).

After that, check whether the current value of θ is greater than 315 (S1005). If the value of θ is not greater than 315, increase the θ value by 15 and repeat steps S1002 to S1005. If the current value of θ is greater than 315 in step S1005, the half transform is terminated.

11 is a diagram illustrating a configuration of an effective parameter detector for preventing detection of overlapping straight lines.

The reconstruction of several straight lines with respect to one straight line causes the image processing apparatus not to search one parameter pair with respect to one edge straight line, and even close parameter pairs similar to the exact value of one parameter. It can be detected as a variable.

In general, the half-transformer determines that all pairs of parameters satisfying the condition of more than a predetermined value are edge straight lines. However, the value of the cell around the exact pair of parameters tends to have a very large value under the influence of the exact pair of parameters, and thus several straight lines are detected.

To solve this problem, the present invention proposes a mechanism for retrieving a cell having the largest value in a parameter space of a certain area and determining that the maximum value is larger than a reference value as a valid parameter pair.

As shown in FIG. 11, the valid parameter detector 140 may include first and second maximum calculators 141 and 143, a plurality of registers 142, and a reference value comparator 144.

The memory 170 shown in FIG. 11 is composed of 15 partition parameter memories, and each partition parameter cell 172 outputs values of cells from address 0 to address 799. That is, the effective parameter detector 140 of FIG. 11 is applied when using the half transform described with reference to FIG. 6. However, the method may be applied even when using the half transform of FIG. 5.

The value output from the split parameter cell 172 is input to the first maximum calculator 141, which calculates the largest maximum value therein. Here, the first maximum calculator 141 corresponds to the 15to1 maximum calculator because the first maximum calculator 141 determines the largest value among the cell values input from the 15 divisional parameter memories.

The maximum value obtained by the first maximum calculator 141 is the largest value among 15 consecutive θ having the same p. This means a column having a maximum value in the parameter space of FIG. 4.

The maximum value obtained is input to C1, the leftmost register, among 15 registers 141. The plurality of registers 141 all store cell values of a column having a maximum value in each row and (ρ, θ) information, which is coordinate information of a corresponding parameter.

The second maximum calculator 143 compares the column maximum values stored in the 15 registers 141 and finds the largest value, that is, the region maximum value. Through operation of the first and second maximum value calculators, the largest value among the predetermined areas 15 × 15 and a parameter pair having the value can be detected.

The reference value comparator 144 compares the searched region maximum value with a predetermined reference value to determine whether the retrieved parameter pair is a valid linear parameter pair.

Meanwhile, C1 of the plurality of registers 142 receives the latest column maximum value, and registers C2 to C14 shift information stored in the next register to the next register. The final C15 register discards the information it has stored. That is, the plurality of registers 142 have a structure similar to a shift register.

12 is a view for explaining matching between the contour image and the infinite linear image according to the present invention.

Image A of FIG. 12 illustrates a contour image stored in the contour image storage unit 171 of the image processing apparatus 100 according to the present invention. In addition, B of FIG. 12 corresponds to an infinite linear image restored by the image processing apparatus 100 using an effective linear parameter obtained from the contour image. That is, the infinite linear image is generated by performing an inverse Hough equation on the effective linear parameter obtained from the contour image.

The matching unit 150 of the image processing apparatus 100 matches the contour image with the infinite linear image reconstructed by the effective linear parameter, that is, the A and B images of FIG. 12. The matching unit 150 acquires a finite straight line image (C of FIG. 12) by selecting only a common pixel among the two images.

In addition, the matching unit 150 stores, for each valid linear parameter pair, in the memory unit 170 information about the starting point and the ending point of the finite linear image section of the infinite straight line restored by the valid linear parameter. . In summary, the image processing apparatus 100 stores the effective linear parameter values ρ and θ and information α and β about the start and end points thereof together in the memory unit 170.

13 is a diagram illustrating an example of matching a contour image and an infinite linear image.

One method of matching the contour image with the infinite straight line image is to calculate the pixel on which the straight line B of FIG. 12 is drawn based on each effective linear parameter pair, and whether the contour image of FIG. 12A exists in the drawn pixel. You can use the verification method.

For example, assume that an infinite straight line restored to the effective linear parameter pair ρ1 and θ1 in FIG. 13 is L1. The matching unit 150 may calculate a plurality of pixels in which an infinite straight line L1 is drawn.

In addition, the matching unit 150 checks whether the contour image of FIG. 12A exists in each of the plurality of pixels. If the contour image of FIG. 12A exists in an arbitrary pixel belonging to an infinite straight line L1, the pixel corresponding to the coordinate is considered to constitute a finite straight line image. On the contrary, even if an arbitrary pixel belongs to an infinite straight line L1, if the contour image of FIG. 12A does not exist in the pixel, the pixel does not correspond to a point forming a finite straight line image.

As such, the matching unit 150 may obtain pixels constituting the line segments S1 and S2 in which the contour image exists among the infinite straight lines L1. In addition, the matching unit 150 may obtain the coordinates of the start points A1 and A2 and the end points B1 and B2 of S1 and S2. The matching unit 150 stores coordinate information of the start points A1 and A2 and the end points B1 and B2 in the memory unit 170 with respect to the infinite straight line L1.

The output unit 160 may draw an infinite straight line L1 using the information of the effective linear parameters ρ1 and θ1 stored in the memory unit 170 later, and the starting points A1 and A2 stored for the infinite straight line L1. It is possible to restore the line segments S1 and S2 using the coordinates of the end points B1 and B2.

14 is a view showing an image processed by the image processing apparatus according to the present invention.

14A corresponds to the original image received from the camera 200 by the image processing apparatus 100 according to the present invention. In the received original image A, a triangular object O1 and two rectangular objects O2 and O3 are represented.

14B illustrates a contour image or an edge image generated by the edge generator 120 of the image processing apparatus 100 from an original image. As can be seen from the figure, the information inside the object existing in the original image does not exist in the contour image image B, and only the contour information of the object exists.

14C is an image reconstructed by the linear parameters obtained through the half transform and the effective linear parameter detection. In an image reconstructed using only effective linear parameter information, only an infinite straight line may exist in which a line segment constituting each object is extended.

With such an infinite straight line, information about the shape or size of the objects O1, O2, and O3 cannot be accurately extracted. In particular, the right side of the rectangular object O2 and the left side of O3 are recognized as one infinite straight line in the image of FIG.

14D corresponds to an image reconstructed using the line parameters and the start and end point information of the straight line. Since the starting point and the ending point information of the straight line exist, the right side of the rectangular object O2 and the left side of O3 can be divided into different line segments. It is possible to accurately determine the shape or size of the original image through such an image.

Although the present invention has been described in detail through the representative embodiments, those skilled in the art to which the present invention pertains can make various modifications without departing from the scope of the present invention. Will understand. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined by the claims below and equivalents thereof.

1 is a diagram illustrating a configuration of an image processing apparatus according to an embodiment of the present invention.

2 is a diagram illustrating an image processing method according to another exemplary embodiment of the present invention.

3 is a diagram illustrating an example of a straight line for performing a half transform;

4 is a diagram schematically showing a half transform result by an image processing apparatus according to the present invention;

5 is a diagram illustrating an example of an operation method of a half transform performed by an image processing apparatus of the present invention.

6 is a diagram illustrating another example of an operation method of a half transform performed by the image processing apparatus of the present invention.

7 illustrates a hardware configuration of a half converter for performing an operation of half conversion.

8 is a diagram illustrating a structure of a parameter cell in which a half transform result of FIG. 6 is stored;

9 is a diagram illustrating a method of performing a half transform of FIG. 6 by an image processing apparatus according to the present invention;

FIG. 10 is a diagram illustrating a method of performing a half transform of FIG. 6 by an image processing apparatus according to the present invention; FIG.

11 is a diagram illustrating a configuration of an effective parameter detection unit for preventing the detection of overlapping straight lines.

12 is a view for explaining matching between the contour image and the infinite straight line image according to the present invention.

FIG. 13 illustrates an embodiment of matching a contour image with an infinite linear image. FIG.

14 is a view showing an image processed by the image processing apparatus according to the present invention.

<Description of the symbols for the main parts of the drawings>

100: image processing device

110: video input unit

120: edge generation unit

130: half transform unit

140: valid parameter detection unit

150: matching unit

160: output unit

170: memory unit

200: camera

Claims (19)

In the image processing method for detecting the start point and the end point of a straight line, The image processing apparatus may further include generating and storing an edge image by using external image data input from the outside, and detecting a straight line parameter by half-converting the generated contour image; The image processing apparatus matching the stored contour image with an infinite linear image reconstructed by performing an inverse half transform on the detected linear parameter; Obtaining only pixels that match the contour image of the pixels constituting the infinite straight line image; And Checking a start point and an end point of a line segment composed of the matching pixels and obtaining information about the starting point and the end point of the line segment; Detecting the linear parameters, Increasing a value of a parameter cell corresponding to a parameter pair obtained by calculating the following equation for a plurality of edge points constituting the contour image by 1; Wow And determining a parameter pair corresponding to the parameter cell as a linear parameter when the value of the parameter cell is equal to or greater than a predetermined value.

(Where x and y are x, y coordinate values of each edge point, ρ, θ are parameters) The method of claim 1, The storing of the generated contour image may include: And the coordinates of valid pixels constituting the generated contour image. delete The method of claim 1, Acquiring a pixel matching the contour image from the infinite straight image by matching the infinite straight line and the contour image, and checking the starting point and the ending point of the line segment composed of the pixel for each linear parameter. Image processing method. delete The method of claim 1, The step of increasing the value of a parameter cell corresponding to a parameter pair obtained by calculating an equation for a plurality of edge points constituting the contour image by 1, And calculating the equations in parallel with respect to a plurality of parameter pairs. delete The method of claim 1, When the value of the parameter cell is greater than or equal to a predetermined value, determining the parameter pair corresponding to the parameter cell as a linear parameter may include: Searching for a cell having the largest value among a plurality of parameter cells corresponding to a predetermined area among the parameter cells, and determining the valid cell as a valid linear parameter when the found cell value is equal to or greater than a predetermined value. Way. An image processing apparatus for detecting a starting point and an ending point of a straight line, An edge generator which generates an edge image using external image data input from the outside and stores the edge image in a memory; A half transform unit for performing a half transform on pixel coordinate information of the contour image generated by the edge generator to obtain a linear parameter; An effective parameter detector for detecting a valid linear parameter among the half-converted linear parameters; And The inverse half-transformation is performed on the valid linear parameters to match the reconstructed infinite linear image and the stored contour image, and only a pixel matching the contour image is obtained from the pixels constituting the infinite linear image, and then the matching is performed. It includes a matching unit for checking the start and end points of the line segment consisting of the pixels, The half transform unit, It is characterized by increasing the value of the parameter cell corresponding to the parameter obtained by calculating the following equation for a plurality of edge points constituting the contour by 1, The valid parameter detection unit, And determining a parameter corresponding to the parameter cell as an effective linear parameter when the value of each cell of the parameter is equal to or greater than a predetermined value.

(Where x and y are x, y coordinate values of each edge point, ρ, θ are parameters) delete 10. The method of claim 9, The matching unit, Acquiring a pixel matching the contour image from the infinite straight image by matching the infinite straight line and the contour image, and checking a start point and an end point of the line segment composed of the pixel for each linear parameter. Image processing device. delete 10. The method of claim 9, A contour image storage unit which stores coordinates of valid pixels constituting the contour image generated by the edge generator; Wow And a valid linear parameter list for storing valid linear parameters checked by the valid parameter detector. delete 10. The method of claim 9, The half transform unit, And calculating the equations in parallel with respect to a plurality of parameter pairs. delete The method of claim 15, The half transform unit, And a plurality of trigonometric functions and arithmetic operators to calculate the equations in parallel for a plurality of parameter pairs. delete 10. The method of claim 9, The valid parameter detection unit, An image processing apparatus comprising: searching for a cell having the largest value among a plurality of parameter cells corresponding to a predetermined area of a parameter space, and determining that the value is a valid linear parameter when the value of the found cell is larger than a reference value .

Images