Inspection method and inspection system

Abstract

To provide an inspection method and an inspection system that can perform inspection with higher accuracy.SOLUTION: An inspection method includes: a first image acquisition step of acquiring a first image by photographing an inspected position at a first angle with respect to an inspected surface in a state where the inspected position on the inspected surface is irradiated with light; a second image acquisition step of acquiring a second image by photographing the inspected position at a second angle different from the first angle with respect to the inspected surface in the state where the inspected position is irradiated with light; and a difference calculation step of calculating difference between the first image and the second image after normalizing difference in an amount of light reflected at the inspected position due to difference between the first angle and the second angle.SELECTED DRAWING: Figure 7

Description

Technical Field

[0001] The present invention relates to an inspection method and an inspection system.

Background Art

[0002] When inspecting whether there are irregularities on the surface of an object to be inspected, it is not easy to distinguish between irregularities and dirt such as black lines. In this regard, according to the surface inspection apparatus of Patent Document 1, it is said that irregularities and dirt on the surface of thin steel plates and thick steel plates can be distinguished.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of the present invention is to provide an inspection method and an inspection system capable of performing inspection with higher accuracy.

Means for Solving the Problems

[0005] According to one aspect of the present invention, a first image acquisition step of irradiating light on a position to be inspected on a surface to be inspected and capturing the position to be inspected at a first angle with respect to the surface to be inspected to obtain a first image; a second image acquisition step of irradiating light on the position to be inspected and capturing the position to be inspected at a second angle different from the first angle with respect to the surface to be inspected to obtain a second image; and a difference calculation step of calculating a difference between the first image and the second image after normalizing a difference in the amount of light reflected at the position to be inspected due to a difference between the first angle and the second angle. An inspection method is provided.

[0006] The inspection method may include a determination step of determining whether there are irregularities at the inspection position based on the difference between the first image and the second image.

[0007] The first angle may be an angle facing the inspection surface directly.

[0008] In the first image acquisition step and the second image acquisition step, the first image and the second image may be respectively acquired by irradiating the inspection position with near-infrared light and photographing the inspection position through a band-pass filter having a wavelength of the near-infrared light as a pass band.

[0009] In the first image acquisition step and the second image acquisition step, the inspection position may be irradiated with light in a mesh shape.

[0010] In the first image acquisition step and the second image acquisition step, the first image and the second image may be respectively acquired by photographing from a camera mounted on a moving body.

[0011] In the first image acquisition step and the second image acquisition step, the inspection position may be irradiated with light from a light source mounted on a moving body.

[0012] In the first image acquisition step and the second image acquisition step, the inspection position may be irradiated with light from the light source through an optical member mounted on the moving body.

[0013] According to another aspect of the present invention, a light source that irradiates light onto a test position on a test surface, and while the test position is irradiated with light, the test position is photographed at a first angle with respect to the test surface to obtain a first image, and while the test position is irradiated with light, the test position is photographed at a second angle different from the first angle with respect to the test surface to obtain a second image, and a difference calculation unit that calculates a difference between the first image and the second image after normalizing a difference in the amount of light reflected at the test position due to a difference between the first angle and the second angle are provided.

Effect of the Invention

[0014] Since the difference in the amount of light reflected at the test position due to the photographing angle is normalized, inspection can be performed with high accuracy.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3A

Figure 3B

Figure 4A

Figure 4B

Figure 5A

Figure 5B

Figure 6

Figure 7

Figure 8

Figure 9

Embodiments for Carrying Out the Invention

[0016] Hereinafter, embodiments according to the present invention will be specifically described with reference to the drawings.

[0017] The method disclosed in Patent Document 1 described above applies the optical cutting method and identifies the positional deviation of the line of the irradiated light with a camera, but does not consider that the amount of light reflection in an inspection object such as a thin steel plate or a thick steel plate depends on the reflection angle. In addition, in the optical cutting method, it is extremely difficult to detect unevenness smaller than the line width of the irradiated light. Therefore, the inventors of the present application have noticed that the method of Patent Document 1 does not necessarily enable accurate detection of unevenness in the inspection object.

[0018] Therefore, in the present embodiment, considering that the amount of light reflection depends on the reflection angle, the unevenness of the inspection surface can be detected with higher accuracy. First, the principle will be explained.

[0019] FIG. 1 is a diagram schematically showing an ideal diffuse reflection model on a reflection surface. In this specification, an angle based on the normal direction of the reflection surface is referred to as an “angle with respect to the reflection surface”. It is assumed that the amount of reflection does not depend on the incident angle of light.

[0020] As shown in the figure, the amount of reflection in the normal direction is high, and the larger the angle θ with respect to the reflection surface, the lower the amount of reflection. Specifically, when the amount of reflection (light intensity) in the normal direction is I0, the amount of reflection (light intensity) in the direction forming an angle θ with respect to the reflection surface is I0cosθ (Lambert's principle).

[0021] FIG. 2 is a diagram for explaining the outline of the measurement method in the present embodiment. It is assumed that the inspection position in the figure is between point A and point B on the inspection surface T (hereinafter referred to as “inspection position A-B”).

[0022] In an example of this embodiment, while irradiating the inspection position A - B with light, the inspection position A - B is photographed by cameras Cx and Cy to respectively obtain two images IMGx and IMGy. Here, the photographing angles of cameras Cx and Cy with respect to the inspection surface T (the angles formed by the optical axes of cameras Cx and Cy and the normal direction, the same hereinafter) are different from each other. For the sake of simplicity of explanation, the photographing angle of camera Cx is set to 0 degrees (that is, facing the inspection surface T directly), and the photographing angle of camera Cy is set to θ. Then, by comparing two images with different photographing angles after performing the following normalization, unevenness can be detected without detecting black lines.

[0023] In addition, in this embodiment, it is assumed that the distances between cameras Cx, Cy and the inspection surface T are about 10 m, and the field angles of cameras Cx, Cy are about 30 degrees. In this case, the distances between an arbitrary position between the inspection positions A - B and cameras Cx, Cy can be regarded as substantially constant.

[0024] FIG. 3A and FIG. 3B are diagrams for explaining the case where there is no unevenness or black line at the inspection position A - B.

[0025] FIG. 3A(a) and FIG. 3A(b) respectively show the brightnesses Ix and Iy of the pixels corresponding to the inspection position A - B in the images IMGx and IMGy from cameras Cx and Cy. When there is no unevenness or black line between the inspection positions A - B, the brightnesses Ix and Iy of the pixels corresponding to the inspection position A - B in both images IMGx and IMGy are constant values. More specifically, for the reason described in FIG. 1, when the brightness Ix in the image IMGx is I0, the brightness Iy in the image IMGy is I0cosθ. Note that the magnitude of the vector indicated by the symbol a in FIG. 3A corresponds to the brightness I0, and the magnitude of the vector indicated by the symbol b corresponds to the brightness I0cosθ.

[0026] FIG. 3B(a) is a reproduction of FIG. 3A(a).

[0027] Figure 3B(b) shows the normalized brightness Iy' of the pixels corresponding to the inspection positions A - B in the image IMGy shown in Figure 3A(b). The normalization here is a process to cancel the difference in the shooting angles of the cameras Cx and Cy. Specifically, Iy' is calculated by dividing the brightness Iy by cosθ. That is, Iy' = Iy / cosθ. As shown in Figure 3A(b), when there are no irregularities or black lines at the inspection positions A - B, since the brightness Iy = I0cosθ, the brightness Iy' = Iy / cosθ = I0.

[0028] In addition, as normalization, a process of normalizing the distortion of the inspection position caused by the difference in the shooting angle may be further performed by means such as affine transformation.

[0029] Figure 3B(c) shows the difference dI between the brightnesses Ix and Iy'. As described above, when there are no irregularities or black lines at the inspection positions A - B, since the brightnesses Ix = Iy' = I0, the difference dI = Ix - Iy' = 0.

[0030] Thus, when there are no irregularities or black lines at the inspection positions A - B, the difference dI becomes 0.

[0031] Figures 4A and 4B are diagrams for explaining the case where there are irregularities at the inspection positions A - B. Figure 4A shows an example where there are no irregularities or black lines between positions A - C and D - B of the inspection positions A - B, but there is a recess (crack) at position C - D. To simplify the explanation, the cross-section of this recess is assumed to be an isosceles triangle with the deepest point at position M, which is the center between positions C - D.

[0032] Figure 4A(a) shows the brightness Ix of the pixels corresponding to the inspection positions A - B in the image IMGx from the camera Cx. Since there are no irregularities or black lines between positions A - C and D - B, the brightness Ix is a constant value I0.

[0033] Between positions C and M, since the shooting angle of camera Cx with respect to the plane including positions C and M deviates from the normal direction (becomes non-zero degrees), the brightness Ix becomes smaller than I0. Similarly, between positions M and D, since the shooting angle of camera Cx with respect to the plane including positions M and D deviates from the normal direction, the brightness Ix becomes smaller than I0. Note that the magnitude of the vector indicated by symbol c in Fig. 4A corresponds to the brightness Ix between positions C and M and between M and D.

[0034] Fig. 4A(b) shows the brightness Iy of the pixels corresponding to the inspection positions A - B in the image IMGy from camera Cy. Since there are no irregularities or black lines between positions A - C and between positions D - B, the brightness Iy is a constant value I0cosθ.

[0035] Between positions C and M, since the shooting angle of camera Cy with respect to the plane including positions C and M approaches the normal direction (becomes smaller than θ), the brightness Iy becomes larger than I0cosθ. Note that the magnitude of the vector indicated by symbol d in Fig. 4A corresponds to the brightness Iy between positions C and M.

[0036] Between positions M and D, since the shooting angle of camera Cy with respect to the plane including positions M and D deviates greatly from the normal direction and becomes almost 90 degrees (the shooting angle becomes almost parallel to the plane including positions M and D), the brightness Iy becomes almost 0.

[0037] Fig. 4B(a) is a reproduction of Fig. 4A(a).

[0038] Fig. 4B(b) shows the normalized brightness Iy’ of the pixels corresponding to the inspection positions A - B in the image IMGy shown in Fig. 4A(b). As shown in the figure, since there are no irregularities or black lines between positions A - C and D - B among the inspection positions A - B, the brightness Iy’ = Iy / cosθ = I0. Among the inspection positions A - B, between positions C and M, since the brightness Iy > I0cosθ, Iy’ > I0. Among the inspection positions A - B, between positions M and D, since the brightness Iy < I0cosθ, Iy’ < I0. In any case, between positions C and D, the brightness Iy’ ≠ I0.

[0039] Figure 4B(c) shows the difference dI between the brightnesses Ix and Iy'. As described above, between positions A - C and D - B among the inspection positions A - B, since the brightnesses Ix = Iy' = I0, the difference dI = Ix - Iy' = 0. On the other hand, between positions C - D where there are unevennesses, since the brightness Iy' ≠ I0, the difference dI ≠ 0.

[0040] In this way, among the inspection positions A - B, the difference dI becomes 0 at positions without unevennesses or black lines, and the difference dI becomes non - zero at positions with unevennesses. Therefore, based on the difference dI, it is possible to identify the presence of unevennesses and their positions.

[0041] Figures 5A and 5B are diagrams for explaining the case where there is a black line at the inspection position A - B. Figure 5A shows an example where among the inspection positions A - B, there are no unevennesses or black lines between positions A - C and D - B, but there is a black line at position C - D.

[0042] Figure 5A(a) shows the brightness Ix of the pixels corresponding to the inspection positions A - B in the image IMGx from the camera Cx. Since there are no unevennesses or black lines between positions A - C and D - B, the brightness Ix = I0.

[0043] On the other hand, between positions C - D, the reflection amount is reduced by the black line (here it is assumed to be 1 / k). Therefore, the brightness Ix = I0 / k. Note that the magnitude of the vector indicated by the symbol e in Figure 5A corresponds to the brightness Ix (= I0 / k) between positions C - D.

[0044] Figure 5A(b) shows the brightness Iy of the pixels corresponding to the inspection positions A - B in the image IMGy from the camera Cy. Since there are no unevennesses or black lines between positions A - C and D - B, the brightness Iy is a constant value I0cosθ.

[0045] On the other hand, between positions C - D, the reflection amount becomes 1 / k due to the black line. Therefore, the brightness Iy = I0cosθ / k. Note that the magnitude of the vector indicated by the symbol f in Figure 5A corresponds to the brightness Iy (= I0cosθ / k) between positions C - D. FIG. 5B(a) is a reproduction of FIG. 5A(a).

[0046] FIG. 5B(b) shows the normalized brightness Iy’ of the pixels corresponding to the inspection positions A - B in the image IMGy shown in FIG. 5A(b). As shown in this figure, since there are no unevenness or black lines between positions A - C and D - B among the inspection positions A - B, the brightness Iy’ = Iy / cosθ = I0. Among the inspection positions A - B, between positions C - D, since the brightness Iy = I0cosθ / k, then Iy’ = I0 / k.

[0047] FIG. 5B(c) shows the difference dI between the brightnesses Ix and Iy’. As described above, among the inspection positions A - B, between positions A - C and D - B, since the brightnesses Ix = Iy’ = I0, the difference dI = Ix - Iy’ = 0. On the other hand, between positions C - D where there are black lines, since the brightnesses Ix = Iy’ = I0 / k, the difference dI = 0.

[0048] Thus, in the inspection positions A - B, whether there are no unevenness or black lines or there are black lines, in any case, the difference dI is 0. Therefore, the black lines are not detected.

[0049] As described above with reference to FIGS. 3A to 5B, in this embodiment, two images of the inspection position A - B on the inspection surface T taken from two angles are compared after normalizing the difference in the amount of light reflection caused by the difference in angles. Thereby, unevenness can be accurately detected without detecting black lines.

[0050] FIG. 6 is a block diagram showing the schematic configuration of an inspection system according to an embodiment.

[0051] In addition to the cameras Cx and Cy described above, the inspection system may include a light source LS. The light source LS is, for example, a laser, which irradiates light at the inspection position on the surface to be inspected. There is no particular limitation on the type of light, and it may be visible light, but near-infrared light is desirable. This is especially the case when performing inspections outdoors because it is less affected by external light such as sunlight. In this case, it is desirable for the cameras Cx and Cy to perform imaging through band-pass filters Fx and Fy having a passband of the wavelength of near-infrared light.

[0052] As an example of light irradiation, the inspection system may include an optical member OM, and the light from the light source LS may be irradiated to the inspection position through the optical member OM. The optical member OM is, for example, a mirror such as a polygon mirror. The laser light from the light source LS is collimated. Therefore, when using an optical member OM such as a polygon mirror, the spread of light is very small at the distance from the light source LS to the inspection position, and the attenuation of light until it reaches the inspection position is negligible.

[0053] As another example of light irradiation, the optical member OM is a lens such as a rod lens, a cylindrical lens, or a Powell lens, and the light from the light source LS is spread through the optical member OM so that a line beam may be irradiated to the inspection position. Even if the light from the light source LS is attenuated, inspection is possible by intentionally spreading the light. By designing the lens system so that the intensity distribution of the line beam is constant, light of a constant intensity is irradiated at any point at the inspection position.

[0054] Also, the shape of the irradiated light is not limited to a line shape, and may be, for example, a mesh shape (lattice shape) composed of a plurality of lines (vertical lines and horizontal lines) orthogonal to each other. By using a mesh shape, more positions can be inspected in a single imaging.

[0055] If there is unevenness in the intensity of the light irradiated at the inspection position, the influence of the intensity unevenness can be suppressed by acquiring the intensity distribution in advance and performing calibration.

[0056] The inspection system includes a moving body 1, and part or all of the cameras Cx, Cy, the light source LS, and the optical member OM may be mounted on the moving body 1. By fixing the cameras Cx, Cy to the moving body 1, the shooting angle can be kept constant. In this case, the cameras Cx, Cy are preferably arranged such that their optical axes are non-parallel to each other. There is no particular limitation on the moving body 1. For example, when the inspection target is a road, a vehicle is suitable, and when the inspection target is a blade of a wind power generation device, a drone is suitable.

[0057] In addition, the inspection system includes an inspection device 2. The inspection device 2 includes an image acquisition unit 21, a normalization unit 22, a difference calculation unit 23, a determination unit 24, and a control unit 25. Part or all of these units may be implemented in hardware, or may be realized by a processor executing a predetermined program. Also, each unit of the inspection device 2 may be provided in one device, or may be distributed and provided in a plurality of devices.

[0058] The image acquisition unit 21 acquires an image IMGx obtained by shooting with the camera Cx and an image IMGy obtained by shooting with the camera Cy. The images IMGx, IMGy may be displayed on a display (not shown).

[0059] The normalization unit 22 normalizes the difference in shooting angles by the cameras Cx, Cy, and generates a normalized image obtained by normalizing the image IMGy.

[0060] Specifically, the normalization unit 22 normalizes the difference in the amount of light reflection on the surface to be inspected caused by the difference in shooting angles. More specifically, when the shooting angle by the camera Cx is θ1 and the shooting angle by the camera Cy is θ2, the normalization unit 22 generates a normalized image by multiplying the brightness of each pixel of the image IMGy by (cosθ1 / cosθ2). The normalized image may be displayed on a display (not shown).

[0061] Further, the normalization unit 22 may normalize the distortion of the inspection position caused by the difference in the shooting angles. Specifically, the normalization unit 22 may perform an affine transformation on the image IMGy to align the shooting directions.

[0062] The difference calculation unit 23 calculates the difference between the image IMGx and the normalized image, and generates a difference image. That is, the difference calculation unit 23 calculates the difference between the image IMGx and the image IMGy after normalizing the difference in the light reflection amount at the inspection position caused by the difference in the shooting angles of the cameras Cx and Cy. The difference image may be displayed on a display (not shown).

[0063] The determination unit 24 determines whether there are irregularities at the inspection position based on the difference image (that is, the difference between the image IMGx and the normalized image). Specifically, the determination unit 24 makes a determination by comparing the brightness in the difference image with a predetermined threshold value. For example, the determination unit 24 determines that there are no irregularities at the positions corresponding to the pixels with brightness less than the threshold value in the inspection position, and determines that there are irregularities at the positions corresponding to the pixels with brightness greater than or equal to the threshold value. By such determination, the presence or absence of irregularities and the positions in the case of having irregularities can be detected. Among the inspection positions, it is determined that there are no irregularities at the positions corresponding to the pixels with brightness less than the threshold value, and it is determined that there are irregularities at the positions corresponding to the pixels with brightness greater than or equal to the threshold value. By such determination, the presence or absence of irregularities and the positions in the case of having irregularities can be detected.

[0064] The control unit 25 controls the cameras Cx and Cy and the light source LS. Specifically, the control unit 25 controls the shooting timing by the cameras Cx and Cy and the light irradiation timing by the light source LS.

[0065] FIG. 7 is a flowchart showing an example of the processing operation of the inspection system. With the light source LS irradiating light to the inspection position, the image acquisition unit 21 acquires the image IMGx obtained by the camera Cx shooting the inspection position at a predetermined shooting angle (for example, a facing angle) (step S21). Also, with the light source LS irradiating light to the inspection position, the image acquisition unit 21 acquires the image IMGy obtained by the camera Cy shooting the inspection position at a predetermined shooting angle (for example, an angle deviated from the facing angle) (step S22). Note that the acquisition of the images IMGx and IMGy may be performed sequentially or simultaneously.

[0066] Subsequently, the normalization unit 22 normalizes the shooting angles by the camera Cx and the shooting angle by the camera Cy (step S23). In particular, the normalization unit 22 normalizes the amount of light reflected at the inspection position due to the difference in shooting angles, normalizes the image IMGy from the camera Cy, and generates a normalized image. Then, the difference calculation unit 23 calculates the difference between the image IMGx from the camera Cx and the normalized image, and generates a difference image (step S24). Based on the difference image, the determination unit 24 determines whether there are irregularities at the inspection position, and if so, determines the position thereof.

[0067] As described above, in the present embodiment, the inspection position is photographed at different shooting angles to generate two images, and after normalizing the difference in shooting angles, the two images are compared. Therefore, the irregularities at the inspection position can be detected with high accuracy.

[0068] Note that the configuration of the inspection system shown in FIG. 6 and the processing operations of the inspection system shown in FIG. 7 are merely examples, and various modifications are possible.

[0069] For example, as shown in FIG. 8, the optical axes of the two cameras Cx and Cy may be arranged to be parallel, and the inspection positions A - B may be located at different positions in the angle of view.

[0070] Also, there may be one camera. That is, with the inspection position irradiated with light, shooting may be performed at a certain shooting angle using the camera, and then, the shooting angle may be changed and shooting may be performed using the same camera. Alternatively, the camera may be fixed and the inspection surface may move. That is, as shown in FIG. 9, when the inspection surface is at a certain location P1, the camera may perform shooting with the inspection position A - B irradiated with light, and then, when the inspection surface moves to another location P2, the camera may perform shooting with the inspection position A - B irradiated with light.

[0071] Furthermore, a part of the processing operations shown in FIG. 7 may be performed by a person. For example, while looking at the difference image generated by the inspection device 2, a person may determine the presence or absence of irregularities and the position of the irregularities.

[0072] The above-described embodiments are described for the purpose of enabling a person having ordinary knowledge in the technical field to which the present invention pertains to implement the present invention. Various modifications of the above embodiments can be naturally made by those skilled in the art, and the technical idea of the present invention can also be applied to other embodiments. Therefore, the present invention should not be limited to the described embodiments, but should be in the broadest scope in accordance with the technical idea defined by the claims.

Explanation of Reference Numerals

[0073] Cx, Cy Camera LS Light Source OM Optical Member Fx, Fy Band-Pass Filter 1 Moving Body 2 Inspection Device 21 Image Acquisition Unit 22 Normalization Unit 23 Difference Calculation Unit 24 Determination Unit 25 Control Unit

Claims

1. A first image acquisition step of acquiring a first image by photographing the inspection position at a first angle with respect to the normal direction of the inspection surface while irradiating the inspection position on the inspection surface with light in a mesh shape; A second image acquisition step of acquiring a second image by photographing the inspection position at a second angle different from the first angle with respect to the normal direction of the inspection surface while irradiating the inspection position on the inspection surface with light in a mesh shape; A difference calculation step of calculating the difference between the first image and the second image after normalizing the difference in the amount of light reflection at the inspection position due to the difference between the first angle and the second angle by multiplying the brightness of each pixel of the second image by (cosθ1 / cosθ2), where the first angle is θ1 and the second angle is θ2. The inspection method comprises the above steps.

2. A first image acquisition step of acquiring a first image by photographing the inspection position at a first angle with respect to the normal direction of the inspection surface through a band-pass filter having a wavelength of the near-infrared light as a pass band while irradiating the inspection position on the inspection surface with near-infrared light; A second image acquisition step of acquiring a second image by photographing the inspection position at a second angle different from the first angle with respect to the normal direction of the inspection surface through a band-pass filter having a wavelength of the near-infrared light as a pass band while irradiating the inspection position on the inspection surface with near-infrared light; A difference calculation step of calculating the difference between the first image and the second image after normalizing the difference in the amount of light reflection at the inspection position due to the difference between the first angle and the second angle by multiplying the brightness of each pixel of the second image by (cosθ1 / cosθ2), where the first angle is θ1 and the second angle is θ2. The inspection method comprises the above steps.

3. A first image acquisition step of acquiring a first image by photographing the inspection position at a first angle with respect to the normal direction of the inspection surface from a camera mounted on a drone while irradiating the inspection position on the inspection surface with light; A second image acquisition step of acquiring a second image by photographing the inspection position at a second angle different from the first angle with respect to the normal direction of the inspection surface from a camera mounted on a drone while irradiating the inspection position on the inspection surface with light; When the first angle is θ1 and the second angle is θ2, the brightness of each pixel of the second image is multiplied by (cosθ1 / cosθ2) to normalize the difference in the amount of light reflected at the inspection position due to the difference between the first angle and the second angle, and then a difference calculation step of calculating the difference between the first image and the second image. The inspection method includes this step.

4. In the first image acquisition step and the second image acquisition step, the inspection method according to claim 3, wherein light is irradiated from a light source mounted on the drone to the inspection position.

5. In the first image acquisition step and the second image acquisition step, the inspection method according to claim 4, wherein light is irradiated from the light source to the inspection position through an optical member mounted on the drone.

6. The inspection method according to any one of claims 1 to 5, further comprising a determination step of determining whether there are irregularities at the inspection position based on the difference between the first image and the second image.

7. The inspection method according to any one of claims 1 to 6, wherein the first angle is an angle facing the inspection surface.

8. A light source that irradiates light in a mesh pattern at the inspection position on the inspection surface, While the inspection position is irradiated with light in a mesh pattern, with respect to the normal direction of the inspection surface One or more cameras that acquire a first image by photographing the inspection position at a first angle and acquire a second image by photographing the inspection position at a second angle different from the first angle with respect to the normal direction of the inspection surface while the inspection position is irradiated with light in a mesh pattern, When the first angle is θ1 and the second angle is θ2, the brightness of each pixel of the second image is multiplied by (cosθ1 / cosθ2) to normalize the difference in the amount of light reflected at the inspection position due to the difference between the first angle and the second angle, and then a difference calculation unit that calculates the difference between the first image and the second image. The inspection system includes this unit.

9. A light source that irradiates near-infrared light at the inspection position on the inspection surface, While irradiating the inspection position with near-infrared light, through a band-pass filter having a wavelength of the near-infrared light as a pass band, the inspection position is photographed at a first angle with respect to the normal direction of the inspection surface to obtain a first image, and while irradiating the inspection position with near-infrared light, through a band-pass filter having a wavelength of the near-infrared light as a pass band, the inspection position is photographed at a second angle different from the first angle with respect to the normal direction of the inspection surface to obtain a second image, and one or more cameras; When the first angle is θ1 and the second angle is θ2, by multiplying the brightness of each pixel of the second image by (cosθ1 / cosθ2), after normalizing the difference in the amount of light reflection at the inspection position due to the difference between the first angle and the second angle, a difference calculation unit that calculates the difference between the first image and the second image. A inspection system comprising: **Claim 10** A light source that irradiates light on an inspection position on an inspection surface; While irradiating the inspection position with light, the inspection position is photographed at a first angle with respect to the normal direction of the inspection surface to obtain a first image, and while irradiating the inspection position with light, the inspection position is photographed at a second angle different from the first angle with respect to the normal direction of the inspection surface to obtain a second image, and one or more cameras mounted on a drone; When the first angle is θ1 and the second angle is θ2, by multiplying the brightness of each pixel of the second image by (cosθ1 / cosθ2), after normalizing the difference in the amount of light reflection at the inspection position due to the difference between the first angle and the second angle, a difference calculation unit that calculates the difference between the first image and the second image. A inspection system comprising:

Images