Line concentration analysis image generation method and line concentration analysis image generation device

Abstract

To shorten generation time of a line-convergence index analysis image by reducing a computational load when evaluating a convergence index, which is the degree of convergence of linear patterns to one point, from a captured image using a line-convergence index filter.SOLUTION: A line-convergence index analysis image generation method includes: a matching degree calculation map generation step of generating a plurality of matching degree calculation maps in which a matching degree between a maximum gradient vector obtained by applying a differential filter to each pixel of a captured image and an orthogonal vector that is orthogonal to a vector of interest passing through a pixel of interest and has a direction to the side of the vector of interest is calculated for each vector of interest and for each search area set on both sides of the vector of interest while using the pixel of interest as a reference; and a moving average map generation step of generating a plurality of moving average maps in which an average value of the matching degrees of each pixel in the search area is calculated for each pixel of interest by applying a moving average filter corresponding to the search area to each matching degree calculation map in the direction of the vector of interest, wherein a line-convergence index analysis image is generated by using the moving average maps.SELECTED DRAWING: Figure 9

Description

Technical Field

[0001] The present invention relates to a line concentration analysis image generation method and a line concentration analysis image generation apparatus that can reduce the calculation load and shorten the generation time of a line concentration analysis image when evaluating the concentration degree, which is the degree of concentration of a linear pattern to a single point in a captured image, using a line concentration filter.

Background Art

[0002] Conventionally, there is a method that uses a line concentration filter to evaluate the concentration degree, which is the degree of concentration of a linear pattern to a single point in a captured image. In particular, in the reading of medical images, the evaluation of the concentration degree of lines or vectors often has important significance. Using a line concentration filter, it is possible to detect the fold pattern of the gastric wall, the retraction of blood vessels around a malignant tumor, spiculation, etc. in a double-contrast stomach X-ray image.

[0003] Patent Document 1 discloses a method in which luminance gradient vectors are measured for a plurality of measurement points on an image, the concentration degree of the luminance gradient vectors is calculated, and when the line concentration reaches a maximum, it is determined that there is line information.

[0004] Patent Document 2 discloses a method in which a density gradient vector is obtained from the light and dark density of a transmission image of a resin molded body, and by emphasizing the light and dark density difference between a cavity included in the resin molded body and the peripheral part using a concentration filter, the cavity is discriminated.

[0005] Patent Document 3 discloses a method in which, in an image of a specimen such as a conductor wafer, a region having a luminance equal to or higher than a predetermined threshold value is extracted as a defect region, the luminance gradient vector of the extracted defect region is calculated, and the defect is classified into a foreign matter defect, a non-uniformity defect, etc. based on the component amount of the luminance gradient vector.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0007] By the way, when generating a line concentration analysis image for a captured image using a line concentration filter that evaluates the degree of concentration of a linear pattern to a single point in the captured image, there is a problem that the computational load is large and it takes a long time to obtain the line concentration analysis image.

[0008] The present invention has been made to solve the above problems, and when evaluating the degree of concentration, which is the degree of concentration of a linear pattern to a single point from a captured image using a line concentration filter, it reduces the computational load and shortens the generation time of the line concentration analysis image. An object of the present invention is to provide a line concentration analysis image generation method and a line concentration analysis image generation apparatus.

Means for Solving the Problems

[0009] In order to solve the above-described problems and achieve the object, the present invention provides a method including: a degree-of-match calculation map generation step of generating a plurality of degree-of-match calculation maps in which the degree of match between a maximum gradient vector obtained by applying a differential filter to each pixel of a captured image and an orthogonal vector that is orthogonal to a target vector passing through a target pixel and has a direction on the target vector side is calculated for each target vector and for each search region set on both sides of the target vector with respect to the target pixel; a moving average map generation step of applying a moving average filter corresponding to the search region to each degree-of-match calculation map in the direction of the target vector to calculate an average value of the degrees of match of each pixel in the search region for each target pixel, thereby generating a plurality of moving average maps; and a line concentration analysis image generation step of referring to the moving average maps for each target vector and each search region, averaging the average values of the degrees of match of each search region centered on the target pixel, setting the highest degree-of-match value among the average degrees of match of each target vector as an analysis result value of the target pixel, and generating a line concentration analysis image in which the analysis result value is associated with each pixel of the captured image.

[0010] Further, in the present invention, in the above invention, the degree-of-match calculation map generation step generates a degree-of-match calculation map for each target vector corresponding to one search region, and generates a map obtained by inverting the sign of each generated degree-of-match calculation map as a degree-of-match calculation map for each target vector corresponding to the other search region.

[0011] Further, in the present invention, in the above invention, the moving average map generation step rotates the degree-of-match calculation map so that the direction of the target vector of the degree-of-match calculation map becomes a reference movement direction of the moving average filter, applies the moving average filter in the reference movement direction on the rotated degree-of-match calculation map, and then rotates the degree-of-match calculation map in the reverse direction to return it to the original position.

[0012] Further, in the present invention, in the above invention, the moving average map is associated with the position of the target pixel with respect to the captured image.

[0013] Further, in the present invention, in the above invention, the orthogonal vector is characterized in that it is in a direction away from the target vector.

[0014] Moreover, the present invention includes an image acquisition unit that acquires an image to be processed as a captured image, a maximum gradient vector obtained by applying a differential filter to each pixel of the captured image, and an orthogonal vector that is orthogonal to the target vector passing through the target pixel and has a direction on the target vector side. A plurality of degree-of-match calculation maps are generated for each target vector and for each search region set on both sides of the target vector with respect to the target pixel, and for each degree-of-match calculation map, a moving average filter corresponding to the search region is applied in the direction of the target vector to calculate the average value of the degrees of match of each pixel in the search region for each target pixel. A plurality of moving average maps are generated in advance. Referring to the moving average maps for each target vector and each search region, among the average degrees of match of each target vector obtained by averaging the average values of the degrees of match of each search region centered on the target pixel, the value with the highest degree of match is used as the analysis result value of the target pixel, and a line concentration analysis image is generated by associating the analysis result value with each pixel of the captured image. A line concentration analysis image generation unit, and a display processing unit that performs a process of displaying and outputting the line concentration analysis image.

Advantages of the Invention

[0015] According to the present invention, when evaluating the degree of concentration, which is the degree of concentration of a linear pattern to a single point, from a captured image using a line concentration filter, the calculation load can be reduced and the generation time of the line concentration analysis image can be shortened.

Brief Description of the Drawings

[0016]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

MODE FOR CARRYING OUT THE INVENTION

[0017] Hereinafter, with reference to the accompanying drawings, a line concentration analysis image generation method and a line concentration analysis image generation apparatus according to the present embodiment will be described.

[0018] <Schematic Configuration> FIG. 1 is a schematic diagram showing the configuration of a line concentration analysis image generation apparatus 1 according to the present embodiment. As shown in FIG. 1, the line concentration analysis image generation apparatus 1 acquires an image of the surface S of the molded part 100 captured by the imaging unit 6 as an imaging image to be processed, applies a line concentration filter to the imaging image to generate a line concentration analysis image, and inspects the presence or absence of defective portions such as flash and sink marks on the surface S. The line concentration analysis image generation apparatus 1 includes an input unit 2, a display unit 3, a storage unit 4, a control unit 5, and an imaging unit 6. Note that the molded part 100 is, for example, a part formed by resin injection molding.

[0019] The input unit 2 is an input interface such as a mouse or a keyboard. The display unit 3 is a display interface such as a liquid crystal display for displaying various information. The storage unit 4 is a storage device such as a hard disk drive or a non-volatile memory. The imaging unit 6 is an imaging device such as a camera.

[0020] The control unit 5 is a control unit that controls the entire line concentration analysis image generation apparatus 1, and includes an image acquisition processing unit 10, a parameter setting unit 11, a line concentration analysis image generation unit 12, and a display processing unit 13. The control unit 5 stores programs corresponding to these functional units in a storage device such as a non-volatile memory or a magnetic disk device, loads these programs into the memory, and executes them with a CPU to execute the corresponding processes.

[0021] The image acquisition processing unit 10 operates the imaging unit 6 to acquire a surface image of the surface S of the molded part 100 as a captured image. Note that the imaging unit 6 and the image acquisition processing unit 10 function as an image acquisition unit.

[0022] The parameter setting unit 11 sets parameters of a line concentration filter including a search area according to the size and unevenness characteristics of defective portions such as tracimas and sink marks generated on the surface S of the molded part 100. The sizes and unevenness characteristics of tracimas and sink marks are different. Therefore, when emphasizing a tracima, a tracima detection mode in which a search area is set according to the size and unevenness characteristics of the tracima may be selected, and when emphasizing a sink mark, a sink mark detection mode in which a search area is set according to the size and unevenness characteristics of the sink mark may be selected.

[0023] The line concentration analysis image generation unit 12 obtains, for each pixel in the search area set based on a target pixel that is each pixel of the captured image, the degree of coincidence between the maximum gradient vector obtained by applying a differential filter to each pixel in the search area and an orthogonal vector that is orthogonal to the target vector passing through the target pixel and has a direction on the target vector side, for each rotation of the target vector around the target pixel, and sets the value with the highest degree of coincidence as the analysis result value of the target pixel, and generates a line concentration analysis image obtained by applying a line concentration filter for obtaining the analysis result value for each pixel of the surface image.

[0024] The display processing unit 13 displays the generated line concentration analysis image on the display unit 3. Note that the display processing unit 13 may superimpose and display the line concentration analysis image on the captured image by an arbitrary operation.

[0025] <Overall Processing of Line Concentration Analysis Image Generation Device> FIG. 2 is a flowchart showing the overall processing procedure by the control unit 5 of the line concentration analysis image generation device 1. As shown in FIG. 2, first, the control unit 5 acquires, as a captured image, a surface image of the surface S of the molded part 100 by the imaging unit 6 (step S110).

[0026] Thereafter, the parameter setting unit 11 sets setting parameters in the line concentration filter (step S120). Then, the line concentration analysis image generation unit 12 performs a generation process of a line concentration analysis image for generating a line concentration analysis image using the line concentration filter (step S130). Then, the display processing unit 13 displays and outputs the generated line concentration analysis image to the display unit 3 (step S140), and this process ends.

[0027] <Outline of Generation of Line Concentration Analysis Image Using Line Concentration Filter> FIG. 3 is an explanatory diagram for explaining the outline of generation of a line concentration analysis image with respect to the captured image D1. Further, FIG. 4 is an explanatory diagram for explaining the rotation of the target vector when generating the line concentration analysis image.

[0028] First, when applying a line concentration filter to the captured image D1, one target pixel 21 on the captured image D1 is specified. Also, a target vector A22 passing through this target pixel 21 is specified. Further, a right search region E R and a left search region E L are set. The right search region E R is a rectangular region determined by a right area width W R away from the target vector A22 and a target length L extending in the direction of the target vector A22. Also, the left search region E L is the right search region ER is set on the opposite side and is the left area width W that is away from the target vector A22 L and is a rectangular area determined by the target length L extending in the direction of the target vector A22. Therefore, the search area E is the right search area E R and the left search area E L and is a rectangular area determined thereby. Note that the right area width W R , the left area width W L is a variable not exceeding the maximum area width Wmax that expands and contracts the width of the search area E, and is a value from 0 to Wmax.

[0029] Then, the line concentration filter takes the partial differential of brightness for each pixel in the right search area E R . Specifically, a Sobel filter is applied. The gradient vector 20 that is the direction of this partial differential value is obtained, and the angle φ of the maximum gradient vector having the largest partial differential value is obtained. Then, the angle difference θ between the angle of the orthogonal vector 23 that is orthogonal to the target vector A22 and has a direction toward the target vector A22 side and the angle φ of the maximum gradient vector is obtained. And for each pixel in the right search area E R , cosθ is calculated. Similarly, cosθ is calculated for each pixel in the left search area E L . Note that the direction of the orthogonal vector 23 in the left search area E L is toward the target vector A22 side. The value of cosθ is a value indicating the degree of coincidence between the orthogonal vector 23 and the maximum gradient vector, and takes a value from 1 to -1. The value of cosθ shown in FIG. 3 is a value close to 1, indicating that the maximum gradient vectors are concentrated toward the target vector A22 side. The maximum gradient vector is a vector that points from the darker side to the brighter side in terms of brightness.

[0030] After that, for the right search area E R and the left search area E L in the search area R, the average value cosθave of cosθ of all pixels is calculated. Here, the right area width W R of the right search area E R and the left area width W L of the left search area E Lis a variable equal to or less than the maximum area width Wmax as described above. For example, when the distance up to the maximum area width Wmax is in 8 steps, the right search area E R and the left search area E L each change to 8 regions. In this case, for the right search area E R the average value cosθave is the maximum value among the average values cosθave of the 8 regions. Similarly, for the left search area E L the average value cosθave is the maximum value among the average values cosθave of the 8 regions. Then, the average value of the average value cosθave of the selected right search area E R and the average value cosθave of the selected left search area E L is obtained as the average degree of coincidence cosθave2 with respect to the target vector A22 specified this time.

[0031] Thereafter, as shown in FIG. 4, the target vector A22 passing through the target pixel 21 is rotated half a turn, for example, in 16 steps around the target pixel 21, and the average degree of coincidence cosθave2 with respect to the target vector A22 at each rotation angle is calculated. Since the rotation is 180 degrees, the rotation angle θr in one step is 11.25 degrees. Then, among the average degrees of coincidence cosθave2 with respect to the target vectors A22 (A22-1 to A22-16) at each rotation angle, the maximum value cosθavemax is calculated as the analysis result value for the target pixel 21. When this analysis result value is large, the brightness of the pixel value of the target pixel 21 becomes a large value. Note that the partial differentiation by the above Sobel filter is performed on the brightness, but it may be the luminance. Also, not limited to the Sobel filter, as long as the direction of the gradient vector 20 can be obtained by a differential filter.

[0032] Then, all the pixels on the captured image D1 are specified as the target pixel 21, and the above analysis result value is calculated. Thereby, the captured image D1 becomes a line concentration analysis image having the analysis result value as a pixel.

[0033] Note that the above captured image D1 is the obtained color image converted into a grayscale intensity image. The imaging unit 6 may directly obtain the captured image D1 as an intensity image.

[0034] <Generation of Moving Average Map> Here, in the present embodiment, for each attention vector A22 of each target pixel by the line concentration filter, the calculation of cosθ indicating the degree of match for each pixel in the search region E and the calculation of the average value cosθave are not performed. In the present embodiment, the value of cosθ indicating the degree of match for each pixel in the search region E with respect to one attention vector A22 is generated in advance as a match degree calculation map associated with the target pixel, and a moving average filter is applied to this match degree calculation map to generate in advance a moving average map indicating the value of the average value cosθave for each target pixel with respect to each attention vector A22, and it is set to be referred to when generating a line concentration analysis image using the line concentration filter. Thereby, since it is not necessary to perform the calculation of overlapping cosθ and the calculation of the average value cosθave, the computational load is reduced, and the generation time of the line concentration analysis image can be shortened. Note that the match degree calculation map and the moving average map are generated for each rotating attention vector A22, but further for the right search region E R and the left search region E L for each, and different right area widths W R and left area widths W L are generated for each. Note that the moving average map does not need to be provided for each of the right search region E R and the left search region E L and can be grouped as one search region E, but since this operation is not a duplicate process, it does not have to be grouped as one search region E.

[0035] <Specific Example of Generation of Moving Average Map> FIG. 5 is an explanatory diagram for explaining a specific example of the generation of the moving average map. Here, one right area width W R and one left area width W LThe generation of the moving average map for the search area E determined by will be described. As shown in FIG. 5, first, a differential image D2 obtained by applying a differential filter to the captured image D1 is generated. The value of the maximum gradient vector is shown for each pixel of this differential image D2. Note that the differential image D2 consists of two images, an X component image and a Y component image, and the maximum gradient vector is a composite vector of the X component and the Y component. Then, using the differential image D2, for each of the right search area E R and the left search area E L and for each of the target vectors A22 (A22―1~A22-16), right match degree calculation maps D3―1~D3-16 and left match degree calculation maps D4―1~D4―16 indicating the value of cosθ, which is the degree of match of each pixel, are generated. The right match degree calculation maps D3―1~D3―16 indicate the degree of match for the right search area E R for each of the target vectors A22-1~A22-15. Also, the left match degree calculation maps D4―1~D4―16 indicate the degree of match for the left search area E L for each of the target vectors A22-1~A22-16.

[0036] Thereafter, for each of the right match degree calculation maps D3―1~D3―16 and each of the left match degree calculation maps D4―1~D4―16, moving average maps D5-1~D5―16, D6-1~D6―16 are generated by applying a moving average filter to the area corresponding to the right search area E R and the left search area E L in the direction of each of the target vectors A22-1~A22-16. The pixel value after applying the moving average filter is the average value cosθave for each target pixel.

[0037] Furthermore, in order to associate the pixels on each of the moving average maps D5-1 to D5-16 and D6-1 to D6-16 with the position of the target pixel on the captured image D1, moving average maps D7 (D7-1 to D7-16) and D8 (D8-1 to D8-16) are generated by shifting the pixel positions. This is because the positions of the target pixel and the pixel indicating the average value cosθave for this target pixel are shifted, and by aligning the positions of the target pixel and the pixel indicating the average value cosθave, it becomes easier to refer to the average value cosθave for the target pixel.

[0038] Note that, for example, when the distance up to the maximum area width Wmax is in 8 steps, the right search area E R and the left search area E L each create moving average maps for 8 search areas in the same manner as described above. Therefore, when the distance up to the maximum area width Wmax is in 8 steps, the number of values generated is the number obtained by multiplying the number of steps on the left and right of the maximum area width Wmax = 16 (8×2) by the number of rotations of the target vector A22 = 16, which is 256.

[0039] <An example of moving average processing> FIG. 6 is a diagram showing an example of moving average processing using a moving average filter. As shown in FIG. 6, when performing a moving average on the right matching degree calculation map D3-8, a moving average filter OP in the same area as the right search area E R is applied to the right matching degree calculation map D3-8 in the direction of the target vector A22-8. The moving average filter OP shown in FIG. 6 is a 3×2 filter with the target length L being 3 pixels and the right area width W R being 2 pixels, and the same weight coefficient "1 / 6" is used for all pixels. Thereby, the moving average map D5-8 is generated.

[0040] <An example of line concentration analysis image> FIG. 7 is a diagram showing an example of a line concentration analysis image. In FIG. 7, for the captured image D11 with high brightness at both the left and right ends and low brightness in the center, when the direction of the gradient vector is set to the bright part direction (the direction of high brightness) and the direction of the orthogonal vector 23 is set to the direction of the target vector A22, a line concentration analysis image in which ridge lines L10 and L11 with high brightness appear at both the left and right ends is obtained. When the direction of the gradient vector for the captured image D11 is set to the dark part direction (the direction of low brightness) and the direction of the orthogonal vector 23 is set to the direction away from the target vector A22, a line concentration analysis image in which a ridge line L12 with low brightness appears in the center is obtained.

[0041] On the other hand, for the captured image D12 with low brightness at both the left and right ends and high brightness in the center, when the direction of the gradient vector is set to the bright part direction (the direction of high brightness) and the direction of the orthogonal vector 23 is set to the direction of the target vector A22, a line concentration analysis image in which a ridge line L13 with high brightness appears in the center is obtained. When the direction of the gradient vector for the captured image D12 is set to the dark part direction (the direction of low brightness) and the direction of the orthogonal vector 23 is set to the direction away from the target vector A22, a line concentration analysis image in which ridge lines L14 and L15 with low brightness appear at both the left and right ends is obtained.

[0042] <Generation processing procedure of line concentration analysis image> FIG. 8 is a flowchart showing the generation processing procedure of the line concentration analysis image by the line concentration analysis image generation unit 12. As shown in FIG. 8, first, the line concentration analysis image generation unit 12 acquires the parameters of the line concentration filter (step S210). The parameters include the search area E, the setting of the area of the target pixel, the rotation angle θr of the target vector A22, the direction of the orthogonal vector 23, the direction of the gradient vector, and the like. Then, a differential image D2 obtained by applying a differential filter to the captured image D1 is generated (step S220). And then, the generation process of the moving average map for each target vector is performed (step S230). The generation process of the moving average map for each target vector will be described later.

[0043] Thereafter, a target pixel 21 on the captured image D1 is specified (step S240), and a target vector A22 passing through the target pixel 21 is specified (step S250). Thereafter, referring to the moving average maps D7 and D8 of the specified target vector A22, the average value cosθave of the two degrees of coincidence within the search region E (E R , E L ) is calculated as the average degree of coincidence cosθave2 of the specified target vector A22 (step S260). Specifically, the average value cosθave of each of the right search region E R and the left search region E L is calculated by using the moving average maps D7 and D8 for a plurality of right search regions E R and left search regions E L with gradually different right area widths W R and left area widths W L . Based on the average value cosθave for each of the plurality of right search regions E R and the plurality of left search regions E L , the maximum value of the average value cosθave of each of the plurality of right search regions E R is determined as one average value cosθave of the right search region E R , and the maximum value of the average value cosθave of each of the plurality of left search regions E R is determined as one average value cosθave of the left search region E L . The value obtained by averaging these two average values cosθave is calculated as the average degree of coincidence cosθave2 of the specified target vector A22. Thus, the average degree of coincidence cosθave2 of one target vector A22 with respect to one target pixel 21 is obtained.

[0044] Thereafter, it is determined whether or not the rotation of the target vector A22 has ended (step S270). If the rotation of the target vector A22 has not ended (step S270: No), the target vector A22 is set to the next rotation angle (step S280), and the process proceeds to step S260 to obtain the average degree of coincidence cosθave2 of the target vector A22 having the next rotation angle with respect to the same target pixel 21.

[0045] On the other hand, if the rotation of the target vector A22 has ended (step S270: Yes), the maximum value cosθavemax of the average degree of coincidence cosθave2 of each target vector A22 is determined as the analysis result value of the target pixel 21 (step S290).

[0046] Thereafter, it is determined whether or not the processing for all the target pixels 21 has been completed (step S300). If the processing for all the target pixels 21 has not been completed (step S300: No), the next target pixel 21 is set (step S310), and the process proceeds to step S240 to repeat the process of determining the analysis result value for the set next target pixel 21.

[0047] On the other hand, if the processing for all the target pixels 21 has been completed (step S300: Yes), a line concentration analysis image composed of the analysis result values of each target pixel 21 is generated (step S320), this process is terminated, and the process returns to step S130.

[0048] <Processing procedure for generating a moving average map for each target vector> FIG. 9 is a flowchart showing a processing procedure for generating a moving average map for each target vector. As shown in FIG. 9, the line concentration analysis image generation unit 12 first determines a target vector A22 (step S410). Further, a search area E (E R , E L ) is determined (step S420). In determining this search area E (E R , E L ), different right area widths W R and left area widths W L are included step by step.

[0049] Thereafter, based on the differential image D2, a degree of coincidence calculation map is generated by calculating the degree of coincidence cosθ based on the angle difference θ between the angle φ of the maximum gradient vector of each pixel in the determined search area E (E R , E L ) and the orthogonal vector 23 (step S430). Thereafter, for the degree of coincidence calculation map, the search area E (E R , E L) Apply a moving average filter corresponding to it in the direction of the target vector A22 to generate a moving average map in which the average value cosθave is calculated (step S440). Further, generate moving average maps D7 and D8 in which the moving average map is associated with the target pixel (step S450).

[0050] After that, it is determined whether there is another search area E (E R , E L ) (step S460). If there is another search area E (E R , E L ) (step S460: Yes), set another search area E (E R , E L ) (step S470), and transfer to step S430. On the other hand, if there is no other search area E (E R , E L ) (step S460: No), it is further determined whether the processing of all the target vectors A22 has been completed (step S470). If the processing of all the target vectors A22 has not been completed (step S480: No), after setting the next target vector A22 (step S490), transfer to step S420. On the other hand, if the processing of all the target vectors A22 has been completed (step S480: Yes), end this processing and return to step S230.

[0051] In this embodiment, a consistency calculation map in which the value of cosθ indicating the consistency of each pixel in the search area E common to one target vector A22 is associated with the target pixel in advance is generated, and a moving average filter is applied to this consistency calculation map to generate in advance a moving average map showing the value of the average value cosθave for each target pixel common to each target vector A22, and it is referred to when generating a line concentration analysis image using a line concentration filter. Therefore, it is not necessary to calculate the overlapping cosθ and the average value cosθave, so the calculation load is reduced and the generation time of the line concentration analysis image can be shortened.

[0052] <Modification Example 1> FIG. 10 is an explanatory diagram for explaining a specific example of generating a moving average map according to Modification 1. In the above-described embodiment, for each search region E R , E L , and for each target vector A22, the right matching degree calculation maps D3-1 to D3-16 and the left matching degree calculation maps D4-1 to D4-16 are generated from the differential image D2. However, in Modification 1, a matching degree calculation map for each target vector corresponding to one search region is generated, and a map obtained by inverting the sign of each generated matching degree calculation map is generated as a matching degree calculation map for each target vector for the other search region.

[0053] Specifically, as shown in FIG. 10, the right matching degree calculation maps D3-1 to D3-16 are generated from the differential image D2, and then, maps obtained by inverting the signs of the right matching degree calculation maps D3-1 to D3-16 are generated as the left matching degree calculation maps D4-1 to D4-16. This is based on the premise that the search regions E R , E L are symmetric with respect to the target vector A22. However, even if the regions of the search regions E R , E L are different, it is considered that there is no significant difference because the average value cosθave is averaged. Note that the same applies to the search regions E R and E L having different right area widths W R , E L . Thereby, generation of one of the matching degree calculation maps can be easily performed.

[0054] <Modification 2> FIG. 11 is an explanatory diagram for explaining a specific example of generating a moving average map according to Modification 2. In this Modification 2, the matching degree calculation map is rotated so that the direction of the target vector A22 of the matching degree calculation map becomes the reference moving direction of the moving average filter, a moving average filter is applied in the reference moving direction on the rotated matching degree calculation map, and then, the matching degree calculation map is rotated back to the original position.

[0055] Specifically, as shown in FIG. 11, the right degree-of-match calculation map D3-4 is rotated by an angle θa formed by the target vector A22-4 and the reference movement direction A of the moving average filter OP, so that the target vector A22-4 and the reference movement direction A of the moving average filter OP are aligned. Thereafter, the moving average filter OP is applied to the rotated right degree-of-match calculation map D3-4' in the reference movement direction A. Thereafter, the moving average map D5-4 to which the moving average filter OP has been applied is rotated counterclockwise by the angle θa to return to its original position. As a result, since the processing of the moving average filter is in the same direction for all the degree-of-match calculation maps, the processing of the moving average filter becomes easier.

[0056] Note that each configuration illustrated in the above embodiments and modification examples is schematically functional, and it is not necessarily physically configured as illustrated. That is, the form of distribution and integration of each device is not limited to that illustrated, and all or part of it can be functionally or physically distributed and integrated in arbitrary units according to various loads, usage situations, and the like.

Industrial Applicability

[0057] The method and apparatus for generating a line concentration analysis image according to the present invention are useful when it is desired to reduce the computational load and shorten the generation time of the line concentration analysis image when evaluating the degree of concentration, which is the degree of concentration of a linear pattern to a single point, from a captured image using a line concentration filter.

Explanation of Reference Numerals

[0058] 1 Line concentration analysis image generation apparatus 2 Input unit 3 Display unit 4 Storage unit 5 Control unit 6 Imaging unit 10 Image acquisition processing unit 11 Parameter setting unit 12 Line concentration analysis image generation unit 13 Display processing unit 20 Gradient vector 21 Pixel of interest 23 Orthogonal Vectors 100 Formed Parts A22, A22―1~A22-15 Vectors of Interest cosθave Average Value cosθave2 Average Consistency cosθavemax Maximum Value D1, D11, D12 Captured Images D2 Differential Image D3―1~D3-16, D3-4´ Right Consistency Calculation Map D4―1~D4-16 Left Consistency Calculation Map D5-1~D5-16, D6-1~D6-16, D7, D7-1~D7-16, D8, D8-1~D8-16 Moving Average Map E Search Region E L Left Search Region E R Right Search Region L10~L15 Ridge Lines OP Moving Average Filter S Surface W L Left Area Width Wmax Maximum Area Width W R Right Area Width θ Angle Difference θa, φ Angles θr Rotation Angle

Claims

1. A degree-of-match calculation map generation step of generating a plurality of degree-of-match calculation maps in which the degree of match between the maximum gradient vector obtained by applying a differential filter to each pixel of the captured image and an orthogonal vector that is orthogonal to the target vector passing through the target pixel and has the direction on the target vector side is calculated for each target vector and for each search region set on both sides of the target vector with respect to the target pixel; A moving average map generation step of applying a moving average filter corresponding to the search region to each degree-of-match calculation map in the direction of the target vector to generate a plurality of moving average maps in which the average value of the degree of match of each pixel in the search region is calculated for each target pixel; A line concentration analysis image generation step of generating a line concentration analysis image in which the highest degree-of-match value among the average degrees of match of each target vector, which are obtained by averaging the average values of the degrees of match of each search region centered on the target pixel with reference to the moving average map for each target vector and each search region, is used as the analysis result value for the target pixel, and the analysis result value is associated with each pixel of the captured image; A line concentration analysis image generation method characterized by including the above steps.

2. The degree-of-match calculation map generation step generates a degree-of-match calculation map for each target vector corresponding to one search region, and generates a map obtained by inverting the sign of each generated degree-of-match calculation map as the degree-of-match calculation map for each target vector for the other search region. The line concentration analysis image generation method according to Claim 1.

3. The moving average map generation step rotates the degree-of-match calculation map so that the direction of the target vector of the degree-of-match calculation map becomes the reference movement direction of the moving average filter, applies the moving average filter in the reference movement direction on the rotated degree-of-match calculation map, and then rotates the degree-of-match calculation map in the reverse direction to return it to its original position. The line concentration analysis image generation method according to Claim 1 or 2.

4. The moving average map is associated with the position of the target pixel with respect to the captured image. The line concentration analysis image generation method according to any one of Claims 1 to 3.

5. The orthogonal vector has a direction away from the target vector. The line concentration analysis image generation method according to any one of Claims 1 to 4.

6. An image acquisition unit that acquires the image to be processed as a captured image; For each pixel of the captured image, the degree of coincidence between the maximum gradient vector obtained by applying a differential filter and an orthogonal vector that is orthogonal to the target vector passing through the target pixel and has the direction on the target vector side is calculated for each target vector and for each search region set on both sides of the target vector with respect to the target pixel, and a plurality of coincidence degree calculation maps are generated. For each coincidence degree calculation map, a moving average filter corresponding to the search region is applied in the direction of the target vector to calculate the average value of the coincidence degrees of each pixel in the search region for each target pixel, and a plurality of moving average maps are generated in advance. Among the average coincidence degrees of each target vector obtained by averaging the average values of the coincidence degrees of each search region centered on the target pixel with reference to the moving average map for each target vector and each search region, the value with the highest degree of coincidence is used as the analysis result value of the target pixel, and a line concentration analysis image is generated by associating the analysis result value with each pixel of the captured image. A line concentration analysis image generation unit; A display processing unit that performs a process of displaying and outputting the line concentration analysis image A line concentration analysis image generation device, characterized by comprising the above.

Images