JPWO2019049316A1 - Deformation detection apparatus and deformation detection method - Google Patents

Abstract

A rectangular setting unit (1031) that sets a plurality of rectangular blocks that divide a visible image in which the surface of the object is photographed, a background luminance value of a visible image in which the surface of the object is photographed, and a rectangle setting unit are set. A candidate rectangle determination unit (103) that determines a deformed candidate rectangle that is a candidate for a deformed rectangle that forms a deformed region among a plurality of rectangular blocks based on the average luminance value in each rectangular block, and a candidate rectangle Among the candidate deformation rectangles determined by the determination unit, a candidate rectangle integration unit (104) for setting a modification candidate region obtained by integrating regions of the modification candidate rectangles adjacent to each other, and a modification candidate set by the candidate rectangle integration unit A candidate contour extracting unit (105) that extracts a contour block that forms the outer periphery of the region, and each contour block is based on the luminance feature amount in each rectangular block around each contour block extracted by the candidate contour extracting unit. Deformation Determining whether the shape, with a candidate contour determination unit and (106) of the region surrounded by the contour block determined as the Deformation region it is determined that the Henjo rectangular.

Description

  The present invention relates to a deformation detection apparatus and a deformation detection method for detecting a deformation location on the surface of an object.

2. Description of the Related Art Conventionally, a technique for detecting a deformed portion on the surface of an object based on image data obtained by imaging the surface of the object is known.
For example, in Patent Document 1, a concrete surface image obtained by imaging a concrete surface is subjected to a strong smoothing process in the lateral direction, and a blacker area is compared with the background for a leak candidate for the image subjected to the smoothing process. Extract the region as a region, specify the brightness value, area, width, and height of the candidate leakage region, extract the region within the specified conditions as the leakage region, fill the region, and close the surrounding region Etc., and a method for detecting a deformed region on the concrete surface that displays the final water leakage region is disclosed.

JP 2012-202859 A

  In the technology as disclosed in Patent Document 1, it is assumed that the water leakage area has a wide width, and since the determination of the water leakage area is determined from the smoothed image, there is a problem that a determination leakage may occur. there were. In addition, since it is determined whether or not it is a leaking area with a fixed value threshold based on the brightness value, area, width, and height, there is a problem that a small leaking area may be missed. It was. In addition, there is a problem in that it is difficult to detect water leakage in a fine area depending on photographing conditions and the like.

  The present invention has been made to solve the above-described problems, and an object thereof is to provide a deformation detection apparatus and a deformation detection method that can detect a deformed area even in a minute area. And

  The deformation detection device according to the present invention includes a rectangular setting unit that sets a plurality of rectangular blocks that divide a visible image in which the surface of the object is photographed, and a background luminance value of the visible image in which the surface of the object is photographed. A candidate rectangle determining unit that determines a deformed candidate rectangle that is a deformed rectangle candidate that forms a deformed region among a plurality of rectangular blocks based on the average luminance value in each rectangular block set by the rectangle setting unit Among the candidate deformation rectangles determined by the candidate rectangle determination unit, a candidate rectangle integration unit that sets a modification candidate region obtained by integrating regions of the modification candidate rectangles adjacent to each other, and a modification set by the candidate rectangle integration unit A candidate contour extracting unit that extracts a contour block that forms the outer periphery of the candidate region, and each contour block is deformed based on a feature value of luminance in each rectangular block around each contour block extracted by the candidate contour extracting unit Is a rectangle UGA determined, in which a region surrounded by the contour blocks is determined that Henjo rectangle and a candidate contour determination unit for determining a Henjo region.

  According to the present invention, it is possible to detect a deformed region even for a fine region.

It is a figure which shows an example of a structure of the deformation | transformation detection apparatus which concerns on Embodiment 1. FIG. 5 is a flowchart for explaining the operation of the deformation detection device in the first embodiment. In Embodiment 1, it is a figure for demonstrating the operation | movement which an image conversion part produces | generates a visible image after conversion. In Embodiment 1, it is a figure for demonstrating the operation | movement which an image conversion part produces | generates a visible image after conversion. In Embodiment 1, it is a figure for demonstrating concretely the operation | movement which a candidate rectangle integration part integrates the area | region of a deformation | transformation candidate rectangle. 9 is a flowchart for explaining in detail the operation of step ST207 by the candidate contour determination unit in the first embodiment. In Embodiment 1, it is a figure for demonstrating the operation | movement which a candidate outline determination part determines whether it is the deformation | transformation rectangle deformed by the water leak about the outline block of a deformation | transformation candidate area | region. It is a figure which shows an example which arranged the average luminance value in an attention outline block, and the average luminance value in each reference block in order of each rectangular block average luminance. It is a figure which shows another example which arranged the average luminance value in an attention outline block, and the average luminance value in each reference block in order of each rectangular block average luminance. In Embodiment 1, it is a figure for demonstrating an example of operation | movement of the distance determination part between contours. FIG. 10 is a diagram for describing an example of an operation of a data generation unit in the first embodiment. 6 is a diagram for describing an example of an image indicated by image data in Embodiment 1. FIG. 13A and 13B are diagrams illustrating an example of a hardware configuration of the deformation detection apparatus according to Embodiment 1.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Embodiment 1 FIG.
The deformation detection device 10 detects a deformation region on the surface of the object based on a visible image obtained by photographing the surface of the object. In the first embodiment, as an example, the object is the inner wall of a concrete tunnel, and the surface of the object is the surface of the inner wall of the tunnel. In the following description, “the inner wall of the tunnel” means “the surface of the inner wall of the tunnel”. Moreover, in Embodiment 1, the deformation detection apparatus 10 shall detect the water leakage area | region in the inner wall of a tunnel as a deformation | transformation area | region.

FIG. 1 is a diagram illustrating an example of a configuration of a deformation detection apparatus 10 according to the first embodiment.
As shown in FIG. 1, the deformation detection device 10 includes an image processing unit 1, a visible image generation unit 2, a data recording unit 3, a data output unit 4, and an image output unit 5.
The visible image generation unit 2 acquires a plurality of captured images obtained by capturing the inner wall of the tunnel from a camera (not shown in FIG. 1) and integrates them into a single image to generate a visible image.
Specifically, the user photographs the inner wall of the tunnel with a camera in the tunnel. At this time, the user shoots the inner wall of the tunnel in a plurality of times. The visible image generation unit 2 acquires a plurality of captured images captured by the user from the camera, performs alignment between images or distortion correction on the acquired plurality of captured images, and A plurality of photographed images are integrated into one visible image obtained by photographing the inner wall of the tunnel. At this time, the visible image generation unit 2 generates a visible image as a grayscale image having 256 gradations. When the range of the inner wall of the tunnel in the visible image is wide, the visible image generating unit 2 generates one visible image for each lining concrete unit of about 10 m × 10 m called “span”, for example.
The visible image generation unit 2 outputs the visible image to the image processing unit 1.

The image processing unit 1 detects a water leakage region (hereinafter referred to as “deformed region”) on the inner wall of the tunnel based on the visible image output from the visible image generation unit 2.
The image processing unit 1 includes an image conversion unit 101, a background luminance value calculation unit 102, a candidate rectangle determination unit 103, a candidate rectangle integration unit 104, a candidate contour extraction unit 105, a candidate contour determination unit 106, The distance determination unit 107, the region integration unit 108, the data generation unit 109, and the image generation unit 110 are included. The candidate rectangle determination unit 103 includes a rectangle setting unit 1031.
The image conversion unit 101 obtains a luminance value distribution (hereinafter referred to as “luminance distribution”) for the visible image output from the visible image generation unit 2. Then, the image conversion unit 101 converts the luminance of the visible image based on the calculated luminance distribution, and generates a converted visible image. The image conversion unit 101 outputs the generated converted visible image to the background luminance value calculation unit 102 and the candidate rectangle determination unit 103.
The background luminance value calculation unit 102 calculates the background luminance value of the converted visible image output from the image conversion unit 101. Specifically, the background luminance value calculation unit 102 creates a luminance value histogram for the converted visible image, and sets the luminance value that is the maximum number of pixels as the background luminance value. The background luminance value calculation unit 102 outputs background luminance value information to the candidate rectangle determination unit 103.

The candidate rectangle determination unit 103 is based on the converted visible image output from the image conversion unit 101 and the background luminance value output from the background luminance value calculation unit 102, and a deformed rectangle candidate that forms a deformed region. The deformation candidate rectangle is determined.
Specifically, first, the rectangle setting unit 1031 of the candidate rectangle determination unit 103 sets a plurality of rectangular blocks that divide the converted visible image.
The candidate rectangle determination unit 103 then uses a plurality of rectangular blocks based on the background luminance value of the converted visible image calculated by the background luminance value calculation unit 102 and the average luminance value in each rectangular block set by the rectangle setting unit 1031. Among them, a rectangular block that is a deformation candidate rectangle is determined. Specifically, when the water leakage area is a deformed area, the candidate rectangle determination unit 103 uses the average luminance value in the rectangular block as a threshold for determining the deformed area (hereinafter referred to as “first threshold”). If it is smaller than (), the rectangular block is determined as a deformable candidate rectangle.

Note that the rectangle setting unit 1031 changes the size of the rectangle of the rectangle block in accordance with the shooting resolution of the shot image obtained by shooting the inner wall of the tunnel by the camera. For example, when the shooting resolution is 0.25 mm per pixel, the rectangular setting unit 1031 divides one block into 8 pixels × 8 lines, and when the shooting resolution is 0.5 mm per pixel, the rectangular setting unit 1031 For example, one block is divided into 4 pixels × 4 lines. The rectangle setting unit 1031 may acquire information on the shooting resolution from the camera.
By making the rectangular size of the rectangular block variable according to the imaging resolution of the captured image, the size of the block allocated to the actual area per unit area can be optimized.
On the other hand, by increasing the size of the rectangular block, there may be a case where the number of blocks having a large luminance gradient or variance, which is a characteristic of deformation, does not satisfy the determination condition for determining a deformed area. Therefore, the accuracy of determination of the deformed region can be increased by reducing the size.
The candidate rectangle determination unit 103 outputs information on the deformed candidate rectangle to the candidate rectangle integration unit 104.

  The candidate rectangle integration unit 104 integrates a plurality of adjacent modification candidate rectangle regions as one region based on the information on the modification candidate rectangle output from the candidate rectangle determination unit 103, and converts the integrated region into a transformation region. It is set as a deformation candidate area that is a candidate for the shape area. The candidate rectangle integration unit 104 outputs information on the deformation candidate area to the candidate contour extraction unit 105.

The candidate contour extraction unit 105 extracts the contour block of the deformation candidate region based on the information on the deformation candidate region output from the candidate rectangle integration unit 104.
In the first embodiment, the outline of a region refers to the outer periphery of the region, and a deformed candidate rectangle that forms the contour of the region is referred to as a contour block. In Embodiment 1, the outline block that forms the outline of the area is a rectangular block in which at least one of the four sides of the rectangular block forms the outline of the area.
At this time, the candidate contour extraction unit 105 gives a contour label to the contour block of the deformation candidate region.
The candidate contour extraction unit 105 uses the extracted contour block information of the deformation candidate region and the information of the contour label attached to the contour block of the deformation candidate region as the deformation candidate contour information. The information is added to the information on the deformation candidate rectangle and output to the candidate contour determination unit 106.

The candidate contour determination unit 106 performs redetermination of the deformation candidate region based on the deformation candidate contour information output from the candidate contour extraction unit 105. Specifically, the candidate contour determination unit 106 determines whether the contour block is a deformed rectangle based on the luminance feature amount of each rectangular block around the contour block for the contour block of the deformed candidate region. Make a decision. Details of the feature amount will be described later.
The candidate contour determination unit 106 determines the region surrounded by the contour block determined to be a deformed rectangle as the deformed region. Further, the candidate contour determination unit 106 determines the deformed candidate rectangle forming the deformed region as the deformed rectangle. At this time, the candidate contour determination unit 106 reassigns the contour label to the contour block determined to be a deformed rectangle.
The candidate contour determination unit 106 uses the information of the contour block of the deformed region and the information of the contour label attached to the contour block of the deformed region as post-determination deformed contour information, and displays the deformed region and the deformed rectangle. It is added to the information and output to the inter-contour distance determination unit 107.

The inter-contour distance determining unit 107 calculates the shortest distance between two deformed regions of the deformed regions based on the post-determination deformed contour information output from the candidate contour determining unit 106, and calculates the shortest distance and the threshold value. (Hereinafter, referred to as “second threshold value”), it is determined whether or not the two deformed regions are to be integrated.
The inter-contour distance determination unit 107 assigns an integration flag to the deformed region determined to be integrated, and integrates the information of the deformed region to which the integration flag is added together with the post-determination deformed contour information and the deformed rectangle information. Output to the unit 108.

The region integration unit 108 integrates the deformed regions to which the integration flag is assigned based on the information on the deformed region output from the inter-contour distance determination unit 107 to obtain a post-integrated deformed region. Re-assign the contour label given to the block.
The region integration unit 108 includes information on the post-integration deformed region, information on the deformed rectangle included in the post-integration deformed region, information on the contour block, and information on the contour label. The data generation unit 109 and the image generation unit To 110.

  The data generation unit 109 generates deformation data based on the post-integration deformed region information, deformed rectangle information, contour block information, and contour label information output from the region integration unit 108. Specifically, the deformation data is data including, for example, the outline of the post-integration deformed area and the label information attached to the outline, the position information of the post-integration deformed area, and the like. The data generation unit 109 outputs the deformed data to the data output unit 4.

The image generation unit 110 uses, as an image, information related to the deformed area based on the post-integrated deformed area information, deformed rectangle information, contour block information, and contour label information output from the area integrating unit 108. Image data for display is generated.
The image generation unit 110 outputs the image data to the image output unit 5.

The data recording unit 3 records the visible image output from the visible image generating unit 2. In addition, the data recording unit 3 is generated in the image processing unit 1 and output in the processing of the image processing unit 1 and the data used in the processing of the image processing unit 1 such as the converted visible image used for the determination of the deformed area. Record intermediate data. The data recording unit 3 records the deformation data and the image data that the image processing unit 1 detects the deformation region and outputs as a final result.
In the first embodiment, the data recording unit 3 is provided in the deformation detection device 10. However, the data recording unit 3 is not limited to this, and the data recording unit 3 is configured to detect the deformation outside the deformation detection device 10. It is good also as a thing provided in the place where the apparatus 10 can be referred.

The data output unit 4 outputs the deformation data output from the data generation unit 109 of the image processing unit 1 to an external recording device or the like (not shown). The data output unit 4 may cause the data recording unit 3 to record the deformation data.
The image output unit 5 outputs the image data output from the image generation unit 110 of the image processing unit 1 to an external display device or the like (not shown). The image output unit 5 may cause the data recording unit 3 to record image data.

Next, the operation of the deformation detection apparatus 10 according to the first embodiment will be described.
FIG. 2 is a flowchart for explaining the operation of the deformation detection apparatus 10 in the first embodiment.

The visible image generation unit 2 acquires a plurality of captured images obtained by capturing the inner wall of the tunnel from the camera, integrates them into one image, and generates a visible image (step ST201). At this time, the visible image generation unit 2 sets the visible image as a grayscale image having 256 gradations.
The visible image generation unit 2 outputs the visible image to the image processing unit 1.

The image conversion unit 101 of the image processing unit 1 converts the luminance in the visible image output from the visible image generation unit 2 in step ST201, and generates a converted visible image (step ST202).
Here, FIG. 3 and FIG. 4 are diagrams for describing an operation in which the image conversion unit 101 generates a converted visible image in the first embodiment.
First, the image conversion unit 101 converts the luminance level of the visible image output from the visible image generation unit 2. Specifically, the image conversion unit 101 obtains an average luminance value for each line of pixels arranged in the horizontal direction on the visible image, and the average luminance value is an intermediate level luminance value of 256 gradations. Correction is performed to shift the luminance value in the visible image so as to be 128. As a result, the image conversion unit 101 reduces unevenness in the luminance value in the vertical direction in the relationship between the imaging light source and the inner wall of the tunnel.
4A is a diagram illustrating an example of a visible image output from the visible image generation unit 2, and FIG. 3A is a diagram illustrating a histogram of luminance values of the visible image as illustrated in FIG. 4A. In FIG. 3, the horizontal axis indicates the luminance value, and the vertical axis indicates the number of pixels at each luminance value.
When the image conversion unit 101 corrects the visible image with the average luminance value in line units set to 128, the corrected visible image becomes an image as shown in FIG. 4B. Moreover, the histogram of the luminance value of the visible image after correction as shown in FIG. 4B is as shown in FIG. 3B.

Subsequently, the image conversion unit 101 obtains a cumulative histogram from the luminance value of the pixel data for the visible image after correction with respect to the visible image so that the slope of the cumulative frequency graph of the cumulative histogram becomes constant. The luminance value of the visible image is converted, and the converted visible image is generated.
The cumulative histogram is obtained by accumulating the number of pixels for each luminance value from the initial luminance value 0 with respect to the histogram of FIG. 3B. The image conversion unit 101 converts the luminance value of the visible image so that the slope of the cumulative pixel count graph of the cumulative histogram is constant.
Specifically, the image conversion unit 101 converts the luminance value of the visible image using the following conversion formula.

x ′ = INT ((hist (x) −xmin) × (L−1) / (1−xmin))
x ′: luminance value after conversion hist (x): cumulative number of pixels with respect to luminance value x before conversion ÷ total number of pixels of visible image after correction xmin: minimum value of hist (x) L: number of gradations (= 256)
INT (): integerization function

FIG. 4C is a diagram illustrating a converted visible image obtained by converting the luminance value of the visible image by the image conversion unit 101, and FIG. 3C is a diagram illustrating a histogram of the converted visible image as illustrated in FIG. 4C.
As described above, the luminance value of the visible image is converted so that the slope of the cumulative frequency graph of the cumulative histogram is constant, and the converted visible image is processed by the illumination condition or the positional relationship between the tunnel inner wall and the camera. This is performed in order to remove the influence of the difference in the brightness value of the photographed image and improve the overall balance of the image.
The image conversion unit 101 outputs the generated converted visible image to the background luminance value calculation unit 102 and the candidate rectangle determination unit 103.

  The background luminance value calculation unit 102 calculates the background luminance value of the converted visible image output from the image conversion unit 101 in step ST202 (step ST203). Specifically, the background luminance value calculation unit 102 creates a histogram of luminance values of the converted visible image, and sets the luminance value that is the maximum number of pixels as the background luminance value. The background luminance value calculation unit 102 outputs background luminance value information to the candidate rectangle determination unit 103.

Candidate rectangle determination unit 103 forms a deformed region based on the converted visible image output from image conversion unit 101 in step ST202 and the background luminance value output from background luminance value calculation unit 102 in step ST203. A deformation candidate rectangle that is a deformed rectangle candidate is determined (step ST204).
Specifically, first, the rectangle setting unit 1031 of the candidate rectangle determining unit 103 divides the converted visible image and sets a plurality of rectangular blocks.
The candidate rectangle determination unit 103 then uses a plurality of rectangles based on the background luminance value of the converted visible image calculated by the background luminance value calculation unit 102 and the average luminance value in each rectangular block set by the rectangle setting unit 1031. Among the blocks, a rectangular block that is a deformation candidate rectangle is determined. Specifically, when the leaked area is a deformed area, the candidate rectangle determining unit 103 determines that the rectangular block is a deformed candidate rectangle when the average luminance value in the rectangular block is smaller than the first threshold value. judge.
The candidate rectangle determination unit 103 sets, for example, a value obtained by subtracting the offset value from the background luminance value (background luminance value− _{Th low1} ) as the first threshold for determining the deformed region. An offset value Th _low1 is set to a value relatively close to the background luminance value. The value may be set by the user in advance.
Note that the rectangular block that is determined by the candidate rectangle determination unit 103 as the deformed candidate rectangle may include deformation due to dirt or uneven color other than water leakage. A rectangular block that has been determined as a deformed candidate rectangle due to a deformation other than water leakage is excluded in the process of the candidate contour determination unit 106 described later.
The candidate rectangle determination unit 103 outputs information on the deformed candidate rectangle to the candidate rectangle integration unit 104.

Candidate rectangle integration section 104 integrates a plurality of adjacent change candidate rectangle areas as one area based on the information of the change candidate rectangle output from candidate rectangle determination section 103 in step ST204, and integrates the areas. Is set as a deformation candidate region (step ST205).
If there is no other deformation candidate rectangle adjacent to the one deformation candidate rectangle for one certain deformation candidate rectangle, the candidate rectangle integration unit 104 converts the one deformation candidate rectangle into the deformation candidate area. And That is, one deformation candidate region is formed by one or more deformation candidate rectangles.

Here, FIG. 5 is a diagram for specifically explaining the operation of the candidate rectangle integration unit 104 integrating regions of deformed candidate rectangles in the first embodiment.
FIG. 5 shows an example of an image showing a deformed candidate rectangle on the converted visible image in which a plurality of rectangular blocks are set.
In FIG. 5, the deformation candidate rectangle is expressed by surrounding a rectangular block with a thick frame.
First, the candidate rectangle integration unit 104 selects one rectangular block among the rectangular blocks determined as the deformed candidate rectangle by the candidate rectangle determination unit 103, and refers to the selected rectangular block (hereinafter referred to as “target block”). 5 is integrated with the area of the rectangular block determined as the deformed candidate rectangle among the eight other rectangular blocks adjacent to the target block.

For example, in FIG. 5, if the position of the target block 501 is represented by the coordinates of X _{i, j} , the candidate rectangle integration unit 104 includes the region of the target block 501 and eight rectangular blocks adjacent to the target block 501. Among the blocks (X _{i−1, j−1 to} X _{i + 1, j + 1} ), the rectangular blocks (X _{i−1, j−1} , X _{i, j−1} , X _{i + 1, j−} ) determined to be deformed candidate rectangles. ₁ , X _{i−1, j} ).
Further, the candidate rectangle integration unit 104 integrates other areas that have already been formed that are adjacent to the area obtained by integrating the area of the current block of interest 501 and the area of the adjacent rectangular block.
For example, in FIG. 5, since there is a region of a rectangular block group that has already been integrated (a rectangular block group of the rectangular block indicated by 502 in FIG. 5), the candidate rectangle integration unit 104 also performs the region of the rectangular block group, Integrate together.
Then, the candidate rectangle integration unit 104 sets the integrated region as a modification candidate region, and outputs information on the modification candidate region to the candidate contour extraction unit 105.

Candidate contour extraction section 105 extracts a contour block that forms the contour of the deformation candidate area based on the information of the deformation candidate area output from candidate rectangle integration section 104 in step ST205 (step ST206).
At this time, the candidate contour extraction unit 105 gives a contour label to the contour block.

For example, if the deformation candidate area has the contents as shown in FIG. 5, the candidate contour extraction unit 105 uses (X _{i−1, j−4} , X _{i + 1,} X _{i, j-4} , Xi _{+ 2, j-4} , Xi _{+ 2, j-3} , Xi _{+ 2, j-2} , Xi _{+ 2, j-1} , Xi _{+ 1, j-1} , Xi _{, j} , Xi _{-1, j} , X _{i-2, j-1} , X _{i-3, j} , X _{i-3, j + 1} , X _{i-2, j + 2} , X _{i-4, j + 1} , X _{i-5, j} , X _{i-4, j-1} , Xi _{-5, j-2} , Xi _{-6, j-2} , Xi _{-6, j-3} , Xi _{-5, j-3} , Xi _{-4, j-2} , X _{i-3, j-2} , Xi _{-2, j-2} , Xi _{-1, j-2} , Xi _{, j-3} ) are extracted, and a contour label is extracted to the extracted contour block. Is granted.
Note that, when there are a plurality of deformation candidate areas, the candidate contour extraction unit 105 assigns different contour labels to the respective deformation candidate areas.
The candidate contour extraction unit 105 uses the extracted contour block information of the deformation candidate region and the information of the contour label attached to the contour block of the deformation candidate region as the deformation candidate contour information. The information is added to the information on the deformation candidate rectangle and output to the candidate contour determination unit 106.

  Candidate contour determination unit 106 performs re-determination of the deformation candidate region based on the deformation candidate contour information output from candidate contour extraction unit 105 in step ST206 (step ST207). Specifically, the candidate contour determination unit 106 determines, with respect to the contour block of the deformation candidate area, the contour block that has been deformed due to water leakage based on the luminance feature amount in each rectangular block around the contour block. It is determined whether or not it is a deformed rectangle that forms a shape area. In step ST207, the candidate change determination area 106 performs the redetermination process for the candidate deformation area by excluding the candidate deformation rectangle due to a change other than leakage from the candidate deformation area, and the change area that has changed due to water leakage and the change area. The purpose is to determine the deformed rectangle included in the shape area.

Here, FIG. 6 is a flowchart for explaining in detail the operation of step ST207 by the candidate contour determination unit 106 in the first embodiment.
The candidate contour determination unit 106 selects one contour block from among the plurality of contour blocks, and 24 rectangular blocks (hereinafter referred to as “target contour blocks”) around the selected contour block (hereinafter referred to as “target contour block”). For the reference block, the average luminance value in each reference block and the value of the variance of the luminance value (hereinafter referred to as “dispersion value”) are calculated (step ST601).
FIG. 7 is a diagram for explaining an operation in which the candidate contour determination unit 106 determines whether or not the contour block of the deformed candidate region is a deformed rectangle deformed due to water leakage in the first embodiment. is there. In FIG. 7, for example, a deformation candidate area as illustrated in FIG. 5 is determined on the converted visible image in which a plurality of rectangular blocks are set, and a contour label is included in the contour block of the determined deformation candidate area. An image to which “1” is assigned is shown.

First, in step ST601, the candidate contour determination unit 106 includes 24 rectangular blocks (X _i−2 ) existing around the target contour block 501 (X _{i, j} ) that is a contour block to which the contour label “1” is assigned. _{, J−2 to} X _{i + 2, j + 2} ) as reference blocks, the average luminance value and the variance value in each reference block are calculated.

Next, candidate contour determination section 106 calculates the number of reference blocks that satisfy the following setting conditions (1) and (2) (step ST602).
(1) Average luminance value of the reference block <background luminance value _{-Th low2 (Th low2 <Th low1} )
(2) variance value of the reference block> background variance _{+ Th variance}

Th _low2 is an intermediate value between the background luminance value and Th _low1 . For example, if the background luminance value is 128 and Th _low1 is 86, Th _low2 is 68, 50, or the like.
The background variance value is the number of rectangular blocks having a variance value in the divided range when the set range is divided as a unit in the histogram of variance values of all rectangular blocks in the converted visible image. The intermediate value of the maximum range. For example, if the set range is 50 in the histogram of variance values for all rectangular blocks in the converted visualized image, the background variance value is 255 when the number of rectangular blocks having variance values 200 to 249 is maximized. And
Also, Th _variance is previously set based on the variance value of the rectangular block corresponding to the background area of a sample of some of the visible image.

Candidate contour determination section 106 determines whether or not the number of reference blocks calculated in step ST602 satisfies the following determination conditions (3) and (4) (step ST603).

(3) Number of reference blocks _satisfying the setting condition of (1) ≦ Th _binarize1
(4) Number of reference blocks _satisfying the setting condition of (2)> Th _binarize2

Here, for example, Th _binarize1 = 8 and Th _binarize2 = 8. Note that the values of Th _binarize1 and Th _binarize2 can be set as appropriate.

Here, FIG. 8 and FIG. 9 are diagrams illustrating an example in which the average luminance value in the target contour block and the average luminance value in each reference block are arranged in the order of the rectangular block average luminance.
In the case as shown in FIG. 8, the above setting condition (3) is satisfied. In the case shown in FIG. 9, the setting condition (3) is not satisfied.

In step ST603, when both the setting conditions of (3) and (4) are satisfied (in the case of “YES” in step ST603), the candidate contour determination unit 106 deforms the target contour block 501 due to water leakage. The deformed rectangle forming the region is determined. (Step ST604).
For example, in the case of FIG. 7, if the number of reference blocks that satisfy the setting condition (1) is “3” and the number of reference blocks that satisfy the setting condition (2) is “10”, Since both of the determination conditions (3) and (4) are satisfied, the candidate contour determination unit 106 determines that the target contour block 501 is a deformed rectangle that forms a deformed region deformed due to water leakage. At this time, the candidate contour determination unit 106 keeps the contour label assigned to the target contour block 501 as it is. That is, the contour label of the target contour block 501 remains “1”.

On the other hand, if any of the determination conditions (3) or (4) is not satisfied in step ST603 (in the case of “NO” in step ST603), the candidate contour determination unit 106 determines that the target contour block 501 is changed due to water leakage. Is determined not to be a deformed rectangle of the deformed region, the information on the contour label attached to the target contour block 501 is deleted, and the target contour block 501 is excluded from the candidate transform region (step ST605). . That is, the candidate contour determination unit 106 excludes the target contour block 501 from the deformed candidate rectangle.
For example, in the case of FIG. 7, if the number of reference blocks satisfying the setting condition (1) is “20”, the number does not satisfy the determination condition of the determination condition (3). The unit 106 deletes the contour label “1” given to the target contour block 501 and excludes the target contour block 501 from the deformation candidate region.
When the water leakage area is a deformed area, the deformed area generally has a sharp brightness gradient in the contour block, and the brightness gradient may be gentle in the contour block other than the water leak due to a stain or the like. Many. The candidate contour determination unit 106 calculates a luminance gradient in each reference block around the target contour block, and the number of reference blocks having a gentle luminance gradient (satisfying the setting condition (1)) is large (the determination condition (3)). Is not satisfied), it is determined that the target contour block is not a deformed region.
In addition, when the water leakage area is a deformed area, the deformed area generally tends to have a larger dispersion of luminance values than the background. The candidate contour determination unit 106 calculates the dispersion value of each reference block around the target contour block, and the number of reference blocks having a large dispersion of the reference block (satisfying the setting condition (2)) is small (the above-mentioned If the determination condition (4) is not satisfied), it is determined that the target contour block is not a deformed region.
As described above, when the water leakage area is a deformed area, the deformed area generally has a luminance value lower than the background luminance value and tends to be a larger variance value than the background variance value. The determination unit 106 makes a determination based on the determination condition (3) and the determination condition (4), and if there are a number of reference blocks including the reference block that correspond to the determination condition (3) and the determination condition (4) equal to or greater than the threshold value. The attention outline block is determined to have a high possibility of a deformed area.

Candidate contour determination section 106 performs the above processing for all contour blocks, and determines whether there is a deformed candidate rectangle that has newly become a contour block in the deformed candidate region (step ST606).
In step ST606, when there is a deformed candidate rectangle that has newly become a contour block (in the case of “YES” in step ST606), the candidate contour determination unit 106 sets the new contour block as a target contour block in steps ST601 to ST601. The process of step ST605 is repeated.
Candidate contour determination section 106 repeats the processing from step ST601 to step ST605 until there is no deformed candidate rectangle that has newly become a contour block.
For example, as shown in FIG. 7, contour blocks (X _{i−4, j−3} , X _{i−4, j−1} , X _{i−5, j} , X _{i−4, j + 1} , X _{i−3, j + 1)} , And X _{i−3, j + 2} ) are excluded from the deformation candidate rectangles of the deformation candidate area.
In this case, the transformation candidate rectangles (X _{i−3, j−1} and X _{i−4, j} ) are newly contour blocks. Therefore, the candidate contour determination unit 106 performs the processing from step ST601 to step ST605 for new contour blocks (X _{i-3, j-1} and X _{i-4, j} ).

In step ST606, when there is no deformation candidate rectangle that has newly become a contour block (in the case of “NO” in step ST606), that is, a deformation that forms a deformed region that has been deformed due to water leakage in all the contour blocks. When the determination of whether or not the shape is rectangular is completed, the candidate contour determination unit 106 reassigns the contour label to the contour block in the deformed candidate region (step ST607). Specifically, when the candidate contour determination unit 106 divides a plurality of deformation candidate regions surrounded by the contour blocks to which the candidate contour extraction unit 105 has assigned the same contour label in step ST206, the candidate contour determination unit 106 separates the candidate candidates. Another contour label is assigned to the contour block of the region.
For example, in FIG. 7, the modification candidate rectangles included in the modification candidate areas indicated by 701 and 702 originally belong to one modification candidate area, and in step ST206, the contour block of the modification candidate area includes An outline label of “1” was given (see FIG. 5). However, as a result of performing processing for determining whether the contour block is a deformed rectangle of the deformed region that has been deformed due to water leakage, the candidate contour determining unit 106 determines that the one deformed candidate region has the deformed state indicated by 701. It was divided into a candidate area and a deformable candidate area indicated by 702.
Therefore, the candidate contour determination unit 106 assigns the contour label “3” to the contour block of the deformable candidate region indicated by 702, for example. Similar to the candidate contour extraction unit 105, the candidate contour determination unit 106 assigns different contour labels to different deformation candidate regions. Now, as shown in FIG. 7, since there is a deformed candidate region 703 in which the contour label “2” is given to the contour block, the candidate contour determining unit 106 includes, for example, the contour block of the deformed candidate region 702 as The contour label “3” that does not overlap with the contour block of another deformation candidate region is assigned.
Note that the candidate contour determination unit 106 may re-assign the specific contour label to the contour block as the contour label as it is for the contour block to which the specific contour label (for example, “1”) has been assigned. In the case where it does not overlap with the contour label of the contour block in the deformable candidate region, the specific contour label may be reassigned to the contour block to which the specific contour label has been assigned. For example, the contour label “1” is re-assigned to the contour block of the deformation candidate region 701 shown in FIG.
And the candidate outline determination part 106 determines with having deformed by water leak, and determines the area | region enclosed by the outline block which gave the outline label again as a deformed area. Further, the candidate contour determination unit 106 determines the deformed candidate rectangle forming the deformed region as the deformed rectangle.
Thereafter, the process returns to step ST204, and the candidate rectangle determination unit 103 determines the value of Th _low1 that is an offset value for the deformed candidate rectangle that becomes the contour block of the deformed candidate area excluding the contour block that has not been determined as the deformed rectangle. Is set to a value smaller than the value at the time of determining the rectangular block that becomes the deformation candidate rectangle for the first time, and the rectangular block that becomes the second deformation candidate rectangle is determined. Then, the subsequent process is performed again, and the candidate contour determination unit 106 integrates the deformed candidate rectangles obtained by integrating the deformed candidate rectangles excluding the contour blocks that are not determined as deformed rectangles in the first determination in step ST207. It is determined whether or not the contour block is a deformed rectangle of the deformed region deformed due to water leakage. The candidate contour determination unit 106 assigns a contour label to the contour block determined to be the deformed rectangle of the deformed region deformed due to water leakage in the second determination, and the region surrounded by the contour block is defined as the deformed region. Determine.
The contour block excluded in the first determination is a rectangular block determined to be a rectangular block deformed due to dirt or uneven color other than water leakage. In the second determination, the determination of the deformed region surrounded by the rectangular block deformed due to water leakage, which is present in the region surrounded by the rectangular block deformed due to dirt or uneven color other than the water leak, is performed. Do.
Thereafter, the determination is repeated until there is no contour block in the deformable candidate area excluding the rectangular block determined to be deformed due to dirt or uneven color.

  The candidate contour determination unit 106 uses the information of the contour block of the deformed region and the information of the contour label attached to the contour block of the deformed region as post-determination deformed contour information, and displays the deformed region and the deformed rectangle. It is added to the information and output to the inter-contour distance determination unit 107.

Returning to the flowchart of FIG.
The inter-contour distance determination unit 107 obtains the shortest distance between two deformed regions of the deformed regions based on the post-determination deformed contour information output from the candidate contour determining unit 106 in step ST207, and determines the shortest distance obtained. Based on the distance and the second threshold, it is determined whether or not the two deformed regions are to be integrated (step ST208).

Here, FIG. 10 is a diagram for explaining an example of the operation of the inter-contour distance determination unit 107 in the first embodiment.
For example, as shown in FIG. 10, it is assumed that four deformed regions are determined by the candidate contour determination unit 106 and contour labels “1” to “4” are assigned to the contour blocks, respectively.
First, the inter-contour distance determination unit 107 obtains the shortest distance between adjacent regions for each deformed region.
For example, in the example of FIG. 10, the inter-contour distance determination unit 107 obtains the shortest distance between adjacent deformed regions as follows.
Minimum distance between the deformed area of the contour label 1 and the deformed area of the contour label 2: 3
Minimum distance between the deformed area of the contour label 1 and the deformed area of the contour label 3: 2
Shortest distance between the deformed area of the contour label 1 and the deformed area of the contour label 4: 10
Shortest distance between the deformed area of the contour label 2 and the deformed area of the contour label 3: 17
Shortest distance between the deformed area of the contour label 2 and the deformed area of the contour label 4: 8
Shortest distance between the deformed area of the contour label 3 and the deformed area of the contour label 4: 21

Next, when the obtained shortest distance between adjacent deformed areas is equal to or less than a preset second threshold Th _connected , the contour distance determining unit 107 determines that the adjacent deformed areas are to be integrated. Then, an integration flag is assigned to the relevant deformed area.
For example, when Th _connected = 8, the inter-contour distance determination unit 107 changes the deformed region of the contour label 1 and the deformed region of the contour label 2, and the deformed region of the contour label 1 and the contour label 3. It is determined that the shape areas are integrated. Then, the inter-contour distance determination unit 107 gives an integration flag to the deformed area determined to be integrated. That is, the integrated flag is assigned to the deformed regions of the contour labels 1 to 3.

  The contour-to-contour distance determination unit 107 outputs the information on the deformed region to which the integration flag has been added to the region integration unit 108 together with the post-determination deformed contour information and the deformed rectangle information.

The region integration unit 108 integrates the deformed regions to which the integration flag is assigned based on the information on the deformed region output from the inter-contour distance determination unit 107 to obtain a post-integrated deformed region. The contour label given to the block is given again (step ST209).
For example, in the example of FIG. 10, the region integration unit 108 integrates the deformed regions of the contour labels 1 to 3 and reassigns the contour label “1” to the integrated contour block of the integrated deformed region. To do.
The region integration unit 108 includes information on the post-integration deformed region, information on the deformed rectangle included in the post-integration deformed region, information on the contour block, and information on the contour label. The data generation unit 109 and the image generation unit To 110.

  Data generation section 109 generates deformation data based on post-integration deformed area information, deformed rectangle information, contour block information, and contour label information output from area integration section 108 in step ST209. (Step ST210). Specifically, the deformation data is data including, for example, the contour of the post-integration deformed region and information on the contour label attached to the contour, the position information of the post-integration deformed region, and the like.

For example, the data generation unit 109 includes all the contour blocks to which the same contour label is assigned for each post-integration deformed region, and the post-integration deforms lines that circumscribe all the contour blocks with the shortest straight line. The contour outline of the region is used. In addition, the data generation unit 109 uses the coordinates of the contour block existing at the end point where the area of the contour block and the lack of convexity as a minimum is the positional information of the post-integration deformed region. For example, in the example of FIG. 10, the data generation unit 109 sets the line indicated by the bold line in FIG. 11 as the post-integration outline 1101 of the post-integration deformed area, and changes the coordinates of the outline block indicated by the white circle 1102 to the post-integration change. Position information.
Note that the coordinates of the contour block are the coordinates of the pixel unit on the converted visible image, and are the coordinates of the upper left, lower left, upper right, and lower right, which are the end points of the contour block including the contour.
The data of the contour encircling line and the positional information of the post-integration deformed area are included in the deformed data.
Then, the data generation unit 109 outputs the deformed data to the data output unit 4.

The data output unit 4 outputs the deformed data output from the data generation unit 109 in step ST210 to an external recording device or the like (not shown) (step ST212). The data output unit 4 may cause the data recording unit 3 to record the deformation data.
In this way, by converting the positional information and the like for the outline of the deformed region after integration into data and recording it, for example, the deformed data recorded in the external recording device or the like by other analyzers or the like can be stored. It can be referred to and used for deformation region analysis processing or the like.

The image generation unit 110, based on the post-integration deformed region information, deformed rectangle information, contour block information, and contour label information output from the region integration unit 108 in step ST209, information on the deformed region. Is generated as an image (step ST211). Specifically, as shown in FIG. 12, the image indicated by the image data is an image in which the contour of the deformed region is superimposed on the converted visible image. That is, the image data is, for example, data for displaying an image as shown in FIG.
The image generation unit 110 outputs the image data to the image output unit 5.

The image output unit 5 outputs the image data output from the image generation unit 110 in step ST211 to an external display device or the like (not shown) (step ST213). The image output unit 5 may cause the data recording unit 3 to record image data.
For example, by displaying information on the deformed area as an image as an image, the user can check the display device and analyze the deformed area.

In the first embodiment described above, the deformation detection device 10 detects the leakage area of the inner wall of the tunnel as the deformation region. However, this is only an example, and the deformation detection device 10 is a concrete detector. A region where precipitates such as white flower appear on the surface can also be detected as a deformed region.
In this case, since the precipitate such as white flower has high luminance, the candidate rectangle determination unit 103 sets an offset value to the background luminance value as the first threshold when determining the deformed candidate rectangle. The added value (background luminance value + Th _high1 ) is used, and when the average luminance value of the rectangular block is larger than the first threshold value, the rectangular block is determined as a deformed candidate rectangle.
In addition, the candidate contour determination unit 106 sets the following as a setting condition (1) when determining whether or not the contour block of the deformed candidate region is a deformed rectangle of the deformed region deformed by the precipitate: Use setting conditions.
Average luminance value of reference block> background luminance value + Th _high2 (Th _high2 > Th _high1 )
In the first embodiment, the deformation detection apparatus 10 includes the image conversion unit 101. However, the image conversion unit 101 is not essential.
For example, in the case where the original image to be detected for deformation is an image that does not have unevenness unlike an image that is supposed to be uneven, such as an image of an inner wall of a tunnel, the image conversion unit 101 may be omitted. The generation process of the visible image after conversion by the image conversion unit 101 is performed as necessary depending on the shooting environment or the situation.

12A and 12B are diagrams illustrating an example of a hardware configuration of the deformation detection apparatus 10 according to the first embodiment.
In the first embodiment of the present invention, the functions of the image processing unit 1, the visible image generation unit 2, the data output unit 4, and the image output unit 5 are realized by the processing circuit 1201. That is, the deformation detection device 10 detects a water leakage area on the inner wall of the tunnel as a deformation area based on a plurality of captured images acquired from the camera, and displays or stores information regarding the detected deformation area as an image image. A processing circuit 1201 for performing control is provided.
The processing circuit 1201 may be dedicated hardware as illustrated in FIG. 12A or a CPU (Central Processing Unit) 1206 that executes a program stored in the memory 1205 as illustrated in FIG. 12B.

  When the processing circuit 1201 is dedicated hardware, the processing circuit 1201 includes, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), and an FPGA (Field-Programmable). Gate Array) or a combination thereof.

  When the processing circuit 1201 is the CPU 1206, each function of the image processing unit 1, the visible image generation unit 2, the data output unit 4, and the image output unit 5 is realized by software, firmware, or a combination of software and firmware. Is done. That is, the image processing unit 1, the visible image generation unit 2, the data output unit 4, and the image output unit 5 are a CPU 1206 that executes programs stored in an HDD (Hard Disk Drive) 1202, a memory 1205, and the like, a system LSI This is realized by a processing circuit such as (Large-Scale Integration). The program stored in the HDD 1202, the memory 1205, or the like may cause a computer to execute the procedure or method of the image processing unit 1, the visible image generation unit 2, the data output unit 4, and the image output unit 5. I can say that. Here, the memory 1205 is, for example, a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable Read Only Memory), an EEPROM (Electrically Erasable Memory), or the like. Or volatile semiconductor memory, magnetic disk, flexible disk, optical disk, compact disk, mini disk, DVD (Digital Versatile Disc), and the like.

Note that some of the functions of the image processing unit 1, the visible image generation unit 2, the data output unit 4, and the image output unit 5 are realized by dedicated hardware, and part is realized by software or firmware. You may do it. For example, the function of the image processing unit 1 is realized by a processing circuit 1201 as dedicated hardware, and the processing circuit is stored in the memory 1205 for the visible image generation unit 2, the data output unit 4, and the image output unit 5. The function can be realized by reading and executing the read program.
The data recording unit 3 uses an HDD 1202, for example. This is only an example, and the data recording unit 3 may be constituted by a DVD, a memory 1205, or the like. Further, it may be an HDD or the like connected to the deformation detection apparatus 10 via a network.
In addition, the deformation detection apparatus 10 includes an input interface device 1203 and an output interface device 1204 that communicate with an external device such as a camera, an external recording device, or an external display device.

  As described above, according to the first embodiment, the deformation detection device 10 includes the rectangular setting unit 1031 that sets a plurality of rectangular blocks that divide a visible image obtained by photographing the surface of the object, and the surface of the object. Based on the background luminance value of the visible image in which the image is photographed and the average luminance value in each rectangular block set by the rectangular setting unit 1031, among the plurality of rectangular blocks, the deformed rectangle candidate forming the deformed region A candidate rectangle determining unit 103 that determines a candidate deformation rectangle, and a candidate that sets a candidate deformation region that is obtained by integrating the regions of the candidate deformation rectangles adjacent to each other from the candidate deformation rectangles determined by the candidate rectangle determining unit 103 The rectangle integration unit 104, the candidate outline extraction unit 105 that extracts the outline block that forms the outer periphery of the deformation candidate area set by the candidate rectangle integration unit 104, and the surroundings of each outline block extracted by the candidate outline extraction unit 105 Based on the luminance feature value in each rectangular block, it is determined whether or not each contour block is a deformed rectangle, and the region surrounded by the contour block determined to be a deformed rectangle is determined as the deformed region. A candidate contour determination unit 106 is provided. Thereby, it is possible to detect a deformed region even for a fine region.

  In the present invention, any constituent element of the embodiment can be modified or any constituent element of the embodiment can be omitted within the scope of the invention.

  Since the deformation detection device according to the present invention is configured to be able to detect a deformation region even for a fine region, the deformation detection device is applied to a deformation detection device that detects a deformation location on the surface of an object. be able to.

  DESCRIPTION OF SYMBOLS 1 Image processing part, 2 Visible image generation part, 3 Data recording part, 4 Data output part, 5 Image output part, 10 Deformation detection apparatus, 101 Image conversion part, 102 Background luminance value calculation part, 103 Candidate rectangle determination part, 104 candidate rectangle integration unit, 105 candidate contour extraction unit, 106 candidate contour determination unit, 107 distance between contour determination unit, 108 region integration unit, 109 data generation unit, 110 image generation unit, 1031 rectangle setting unit, 1201 processing circuit, 1202 HDD, 1203 input interface device, 1204 output interface device, 1205 memory, 1206 CPU.

Claims (8)

A rectangular setting unit that sets a plurality of rectangular blocks that divide a visible image in which the surface of the object is captured;
Based on the background luminance value of the visible image obtained by photographing the surface of the object and the average luminance value in each rectangular block set by the rectangular setting unit, a variable that forms a deformed area among the plurality of rectangular blocks is formed. A candidate rectangle determination unit that determines a deformed candidate rectangle that is a candidate for a rectangular shape;
A candidate rectangle integration unit for setting a deformation candidate region obtained by integrating the regions of the deformation candidate rectangles adjacent to each other among the deformation candidate rectangles determined by the candidate rectangle determination unit;
A candidate contour extraction unit that extracts a contour block that forms the outer periphery of the deformed candidate region set by the candidate rectangle integration unit;
Based on the luminance feature amount in each rectangular block around each contour block extracted by the candidate contour extraction unit, it is determined whether each contour block is the deformed rectangle, and the deformed rectangle is A deformation detection apparatus comprising: a candidate contour determination unit that determines a region surrounded by the determined contour block as the deformation region. An image conversion unit that generates a converted visible image obtained by converting luminance in a visible image obtained by photographing the surface of the object;
The rectangle setting unit sets a plurality of rectangular blocks for dividing the converted visible image generated by the image conversion unit,
The candidate rectangle determination unit determines the deformation candidate rectangle based on a background luminance value of the converted visible image generated by the image conversion unit and an average luminance value in each rectangular block set by the rectangle setting unit. The deformation detection device according to claim 1, wherein: The rectangle setting unit
The deformation detection device according to claim 1, wherein the size of the rectangular block is variable according to the resolution of the visible image. Whether the candidate contour determination unit determines the shortest distance between two deformed regions out of the plurality of deformed regions, and whether to integrate the two deformed regions based on the determined shortest distance and a threshold value An inter-contour distance determination unit for determining whether or not
When the inter-contour distance determining unit determines that the two deformed regions are integrated, a region integrating unit that integrates the two deformed regions into a post-integrated deformed region;
A deformation data generation unit that generates deformation data composed of information related to the post-integration deformation region integrated by the region integration unit;
The deformation detection device according to claim 1, further comprising: an image generation unit configured to generate image data based on information related to the post-integration deformed region integrated by the region integration unit. The deformed region is a region deformed due to water leakage,
The candidate rectangle determination unit
2. The deformation according to claim 1, wherein when the average luminance value in the rectangular block is smaller than a threshold value obtained by subtracting an offset value from the background luminance value, the rectangular block is determined as the deformation candidate rectangle. Detection device. The deformed region is a region deformed by precipitates,
The candidate rectangle determination unit
2. The deformation according to claim 1, wherein when the average luminance value in the rectangular block is larger than a threshold obtained by adding an offset value to the background luminance value, the rectangular block is determined as the deformation candidate rectangle. Detection device. The variable according to claim 1, further comprising: a visible image generation unit configured to acquire a plurality of photographed images obtained by photographing the target object from a camera, and combine the plurality of photographed images into one sheet to generate the visible image. Status detector. A step of setting a plurality of rectangular blocks by which the rectangular setting unit divides the visible image in which the surface of the object is captured;
The candidate rectangle determination unit is configured to change a variable among the plurality of rectangular blocks based on a background luminance value of a visible image obtained by photographing the surface of the target object and an average luminance value in each rectangular block set by the rectangle setting unit. Determining a deformed candidate rectangle that is a deformed rectangle candidate forming the deformed region;
A candidate rectangle integration unit setting a transformation candidate region obtained by integrating regions of transformation candidate rectangles adjacent to each other among the transformation candidate rectangles determined by the candidate rectangle determination unit;
A candidate contour extracting unit extracting a contour block forming an outer periphery of the deformed candidate region set by the candidate rectangle integrating unit;
The candidate contour determination unit determines whether each contour block is the deformed rectangle based on the luminance feature amount in each rectangular block around each contour block extracted by the candidate contour extraction unit, A deformation detection method comprising: determining an area surrounded by a contour block determined to be a deformed rectangle as the deformed area.

Images