Recognition processing device, recognition processing method, and recognition processing program

Abstract

To more appropriately detect an object to be recognized in image recognition processing.SOLUTION: A recognition processing device 10 includes: a first acquisition unit 11 for acquiring an image for recognition; a recognition unit 13 for recognizing an object from the image for recognition; a second acquisition unit 12 for acquiring a plurality of correction images which include a range of the recognized object, and is generate by capturing light of wavelengths each of which differs one another; and a correction unit 14 for correcting a range of the object based on a degree of light reflection in the range of the object in each of the plurality of correction images.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a recognition processing apparatus, a recognition processing method, and a recognition processing program.

Background Art

[0002] A technique is known for detecting an object such as a pedestrian from an image obtained by imaging the surroundings of a vehicle using image recognition processing such as pattern matching. When the ground where the object exists is wet and a reflected image of the object reflected on the ground can be seen, a technique for detecting the range where the actual object exists with high accuracy has been proposed (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] There is a need for a new technique for appropriately detecting an object to be recognized in image recognition processing.

[0005] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide a new technique for appropriately detecting an object to be recognized in image recognition processing.

Means for Solving the Problems

[0006] The recognition processing apparatus according to an aspect of the present disclosure includes a first acquisition unit that acquires a recognition image, a recognition unit that recognizes an object from the recognition image, a second acquisition unit that acquires a plurality of correction images generated by imaging light of different wavelengths, each including the range of the recognized object, and a correction unit that corrects the range of the object based on the degree of light reflection in the range of the object in each of the plurality of correction images.

[0007] The recognition processing method according to another aspect of the present disclosure includes a step of acquiring a recognition image, a step of recognizing an object from the recognition image, a step of acquiring a plurality of correction images generated by imaging light of different wavelengths, each including the range of the recognized object, and a step of correcting the range of the object based on the degree of light reflection in the range of the object in each of the plurality of correction images.

[0008] The recognition processing program according to still another aspect of the present disclosure causes a computer to execute a step of acquiring a recognition image, a step of recognizing an object from the recognition image, a step of acquiring a plurality of correction images generated by imaging light of different wavelengths, each including the range of the recognized object, and a step of correcting the range of the object based on the degree of light reflection in the range of the object in each of the plurality of correction images.

Advantages of the Invention

[0009] According to an aspect of the present disclosure, it is possible to provide a new technique for appropriately detecting an object to be recognized in image recognition processing.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Mode for Carrying Out the Invention

[0011] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. Specific numerical values and the like shown in such embodiments are merely examples for facilitating the understanding of the invention, and do not limit the present disclosure unless otherwise specified. In the drawings, elements not directly related to the present disclosure are not shown.

[0012] First Embodiment FIG. 1 is a block diagram schematically showing the functional configuration of a recognition processing apparatus 10 according to the first embodiment. The recognition processing apparatus 10 includes a first acquisition unit 11, a second acquisition unit 12, a recognition unit 13, a correction unit 14, and an output control unit 15. The recognition processing apparatus 10 acquires, for example, an image that may include an object to be recognized such as a pedestrian existing around, and detects the object to be recognized included in the image.

[0013] In the first embodiment, a case where the recognition processing device 10 is installed in a multifunctional utility pole used as road infrastructure such as a smart pole (trademark registered) will be exemplified. The multifunctional utility pole of the first embodiment is installed, for example, on a street and includes an antenna and communication equipment for providing a wireless communication function, lighting equipment for illuminating the street, and a first camera 21 and a second camera 22 for photographing vehicles and pedestrians passing on the road. The recognition processing device 10 of the first embodiment is fixed at a predetermined location. However, it is not limited to this, and the recognition processing device 10 may be mounted on a moving body, for example, or may be mounted on a flying body such as a vehicle or a drone.

[0014] The first camera 21 is provided on the multifunctional utility pole and generates an image of the surroundings of the multifunctional utility pole. The first camera 21 is provided, for example, above the multifunctional utility pole and generates an image with a viewing angle that looks down on the ground where the multifunctional utility pole is installed. The first camera 21 of the first embodiment is a far-infrared camera and generates a thermal image by imaging infrared rays. However, it is not limited to this, and the first camera 21 may be, for example, a visible light camera, and may generate a color image or a monochrome image by imaging visible light rays. The image generated by the first camera 21 of the first embodiment is, for example, a moving image such as 30 frames per second or 60 frames per second, but may also be a still image.

[0015] The second camera 22 is provided above the multifunctional utility pole, similarly to the first camera 21. The second camera 22 generates an image within the same range as the imaging range of the first camera 21. Therefore, when an object to be recognized is recognized from the recognition image by the recognition unit 13, the correction image generated by the second camera 22 will include the range of the recognized object. The second camera 22 generates a plurality of images using light of different wavelengths. The second camera 22 of the first embodiment is a multispectral camera. Without being limited thereto, the second camera 22 may be, for example, a hyperspectral camera, or may be a combination of a plurality of cameras that image light in different wavelength bands. The correction image generated by the second camera 22 of the first embodiment is, for example, a moving image such as 30 frames per second or 60 frames per second, but may also be a still image. The second camera 22 of the first embodiment images visible light belonging to a wavelength band of, for example, 380 nm or more and less than 780 nm to generate a visible light image. Further, the second camera 22 of the first embodiment images short-wave infrared light belonging to a wavelength band of, for example, 1000 nm or more and less than 2500 nm to generate a short-wave infrared light image.

[0016] Each functional block shown in the first embodiment can be realized, for example, by the cooperation of hardware and software. The hardware of the recognition processing device 10 is realized by elements and mechanical devices such as a processor such as a computer's CPU (Central Processing Unit) or GPU (Graphics Processing Unit), and a memory such as a ROM (Read Only Memory) or RAM (Random Access Memory). The software of the recognition processing device 10 is realized by a computer program or the like.

[0017] The first acquisition unit 11 acquires the image generated by the first camera 21 as a recognition image for recognizing an object to be recognized.

[0018] The second acquisition unit 12 acquires the image generated by the second camera 22 as a correction image for correcting the range of the object.

[0019] The second acquisition unit 12 of the first embodiment includes a visible light image acquisition unit 16 and a short-wave infrared light image acquisition unit 17. The visible light image acquisition unit 16 acquires a visible light image supplied from the second camera 22. The short-wave infrared light image acquisition unit 17 acquires a short-wave infrared light image supplied from the second camera 22. The short-wave infrared light image of the first embodiment is an example of the first correction image. The visible light image of the first embodiment is an example of the second correction image.

[0020] The recognition unit 13 detects an object to be recognized in the recognition image acquired by the first acquisition unit 11. Hereinafter, in the first embodiment, a person is assumed as the object to be recognized.

[0021] The recognition unit 13 has, as dictionary data, a person identifier generated by learning a large number of images in which people are photographed. The recognition unit 13 searches within the input image acquired by the first acquisition unit 11 using the person identifier. For example, HOG (Histograms of Oriented Gradients) features can be used for person detection. Note that Haar-like features, LBP (Local Binary Patterns) features, etc. may also be used. When a person exists in the input image, the recognition unit 13 recognizes the person with, for example, a rectangular recognition frame. The recognition frame of the first embodiment is an example of the range of the object.

[0022] Note that the object to be recognized is not limited to a person, and may be, for example, an animal such as a dog or a cat, an oncoming vehicle, a preceding vehicle, a bicycle, a motorcycle, etc. The recognition unit 13 recognizes each object in the recognition image based on the identifier of each object.

[0023] The correction unit 14 corrects the range of the object based on the degree of light reflection calculated for each of the plurality of correction images. The method for correcting the range of the object by the correction unit 14 will be described later.

[0024] The output control unit 15 generates object information regarding the object to be recognized recognized by the recognition unit 13 based on the range of the object to be recognized, and causes the output device 30 to output the object information. The object information can include information regarding the position and distance of the object to be recognized, which is obtained using a known method based on the range of the object to be recognized, for example. The object information may include, for example, whether or not the object to be recognized has been recognized by the recognition unit 13 and the number of recognized objects. When the range of the object is corrected, the output control unit 15 generates object information based on the corrected range of the object. The output device 30 may include a communication device, and may include a wireless communication device that outputs object information such as the position and distance of the object by vehicle-road communication or vehicle-vehicle communication.

[0025] A method for correcting the range of an object according to the first embodiment will be described. FIG. 2 is a diagram schematically showing an example of a recognition image F1 including an object O1 to be recognized. In FIG. 2, due to the reflection image R1 of the object O1 to be recognized in the puddle P1 on the road surface and the accuracy of the dictionary for object recognition learned by machine learning, etc., the recognition frame DE1 extends downward from the range of the actual object O1 in the recognition image F1. That is, the recognition unit 13 misrecognizes a part of the reflection image R1 as the object O1 to be recognized, and cannot accurately frame the range of the object O1 in the recognition frame DE1.

[0026] The correction unit 14 of the first embodiment calculates an index indicating the degree of light reflection by moisture based on the light reflectance at each position in the region within the recognition frame DE1 in each of the visible light image and the short-wave infrared light image. Specifically, the correction unit 14 calculates a normalized difference water index (NDWI) for each position in the region within the recognition frame DE1 in each of the visible light image and the short-wave infrared light image. NDWI is an index indicating the presence of water areas on the ground surface. By using NDWI, it is possible to determine the presence or absence of water areas in the region within the recognition frame DE1. NDWI is obtained by the following formula (1). NDWI = (Red - SWIR) / (Red + SWIR) Equation (1)

[0027] In Equation (1), Red represents the reflectance of the red image composed of the red component in the visible light image supplied from the second camera 22, and SWIR represents the reflectance of the short-wave infrared light image supplied from the second camera 22. The correction unit 14 calculates the reflectances of the red image and the short-wave infrared light image using a known method, and calculates NDWI using Equation (1) based on these reflectances.

[0028] NDWI has a value between -1 and +1. Utilizing the characteristic that in water areas, the reflectance of visible light is large and the reflectance of short-wave infrared light is small, the correction unit 14 of the first embodiment identifies a non-recognition target area where there are non-recognition target objects different from the recognition target object among the ranges of the objects based on NDWI. Specifically, an area where NDWI is equal to or greater than a predetermined threshold value (for example, 0.5) is determined as a water area (i.e., a non-recognition target area). In the example of FIG. 2, it is assumed that the overlapping area DU1 between the recognition frame DE1 and the puddle P1 has NDWI equal to or greater than the predetermined threshold value. In this case, the correction unit 14 determines that the overlapping area DU1 between the recognition frame DE1 and the puddle P1 is a non-recognition target area where NDWI is equal to or greater than the predetermined threshold value, and corrects the recognition frame DE1 to exclude this overlapping area DU1 from the recognition frame DE1.

[0029] FIG. 3 is a diagram schematically showing an example of the recognition image F1 after correcting the recognition frame. The correction unit 14 of the first embodiment performs a masking process so as to cover the overlapping area DU1 where NDWI is equal to or greater than the predetermined threshold value with the mask M1. The correction unit 14 generates a corrected recognition frame DE2 such that the boundary of the recognition frame DE1 on the mask M1 side is at a position in contact with the mask M1. For example, the correction unit 14 generates the corrected recognition frame DE2 by correcting the recognition frame DE1 so that the lower frame of the recognition frame is in contact with the upper end UP1 of the mask M1. Thereby, the range of the object is corrected.

[0030] FIG. 4 is a flowchart showing an example of process S100 of the recognition processing method according to the first embodiment. Process S100 is repeatedly executed, for example, every few milliseconds.

[0031] In step S101, the first acquisition unit 11 acquires a recognition image generated by the first camera 21.

[0032] In step S102, the recognition unit 13 determines whether an object to be recognized is detected in the recognition image. If an object to be recognized is detected (Yes in step S102), process S100 proceeds to step S103. If an object to be recognized is not detected (No in step S102), process S100 ends.

[0033] In step S103, the second acquisition unit 12 acquires a correction image.

[0034] In step S104, the correction unit 14 calculates the NDWI for each position in the area within the recognition frame of the object to be recognized in the correction image.

[0035] In step S105, the correction unit 14 determines whether there is an area where the NDWI is greater than or equal to the threshold within the recognition frame of the object to be recognized. If there is an area where the NDWI is greater than or equal to the threshold (Yes in step S105), process S100 proceeds to step S106. If there is no area where the NDWI is greater than or equal to the threshold (No in step S105), process S100 ends.

[0036] In step S106, the correction unit 14 masks the area where the NDWI within the recognition frame of the object to be recognized is greater than or equal to the threshold.

[0037] In step S107, the correction unit 14 corrects the recognition frame so that the lower frame of the recognition frame touches the upper end of the mask. After step S107, process S100 ends.

[0038] In the first embodiment, the recognition processing apparatus 10 includes a first acquisition unit 11 that acquires a recognition image, a recognition unit 13 that recognizes an object from the recognition image, a second acquisition unit 12 that acquires a plurality of correction images generated by imaging light of different wavelengths including the range of the recognized object, and a correction unit 14 that corrects the range of the object based on the degree of light reflection in the range of the object in each of the plurality of correction images. According to this configuration, since the range of the object can be appropriately corrected based on the degree of light reflection, it is possible to appropriately detect the object to be recognized in the image recognition process.

[0039] In the first embodiment, the second acquisition unit 12 acquires, as the plurality of correction images, a first correction image generated by imaging light having a wavelength in the wavelength band of short-wave infrared light and a second correction image generated by imaging visible light having a wavelength in a wavelength band different from the wavelength band of short-wave infrared light. The correction unit 14 calculates an index indicating the degree of light reflection by moisture based on the light reflectance in the range of the object in each of the first correction image and the second correction image. According to this configuration, since it is possible to suppress false detection due to, for example, the reflection image of the object to be recognized being reflected in the water area on the ground surface, it is possible to appropriately detect the object to be recognized in the image recognition process.

[0040] In the first embodiment, the correction unit 14 identifies, based on the degree of light reflection, a non-recognition target area in the range of the object where there is a non-recognition target object different from the recognition target object, and corrects the range of the object so that the boundary on the non-recognition target area side in the range of the object is at a position in contact with the non-recognition target area. According to this configuration, since the non-recognition target area can be appropriately excluded from the range of the object, it is possible to appropriately detect the object to be recognized in the image recognition process.

[0041] Second Embodiment FIG. 5 is a block diagram schematically showing the functional configuration of the recognition processing apparatus 10 according to the second embodiment. The second embodiment is different from the first embodiment in that the second acquisition unit 12 includes a near-infrared light image acquisition unit 18 instead of the visible light image acquisition unit 16. Hereinafter, the second embodiment will be described centering on the differences from the first embodiment, and the common descriptions will be omitted as appropriate.

[0042] The second acquisition unit 12 according to the second embodiment includes a short-wave infrared light image acquisition unit 17 and a near-infrared light image acquisition unit 18. The second camera 22 according to the second embodiment captures near-infrared light belonging to a wavelength band of, for example, 780 nm or more and less than 1000 nm to generate a near-infrared light image. The near-infrared light image acquisition unit 18 acquires the near-infrared light image from the second camera 22. The near-infrared light image according to the second embodiment is an example of the second correction image.

[0043] A method for correcting the range of the object according to the second embodiment will be described. FIG. 6 is a diagram schematically showing an example of a recognition image F2 including an object O2 to be recognized. In FIG. 6, for example, due to the shadow SH1 of the object O2 to be recognized on the soil SL1 such as sand or asphalt on a sunny day or the accuracy of the dictionary for object recognition learned by machine learning, the recognition frame DE3 extends downward from the range of the actual object O2 in the recognition image F2. That is, the recognition unit 13 misrecognizes a part of the shadow SH1 as the object O2 to be recognized, and the range of the object O2 cannot be accurately framed in the recognition frame DE3.

[0044] The correction unit 14 of the second embodiment calculates an index indicating the degree of light reflection by the soil based on the reflectance of light at each position in the region within the recognition frame DE3 in each of the short-wave infrared light image and the near-infrared light image. Specifically, the correction unit 14 calculates the Normalized Difference Soil Index (NDSI) for each position in the region within the recognition frame DE3 in each of the short-wave infrared light image and the near-infrared light image. NDSI is an index indicating the presence of soil on the ground surface. By using NDSI, it is possible to determine the presence or absence of soil in the region within the recognition frame DE3. NDSI is obtained by the following formula (2). NDSI=(SWIR-NIR) / (SWIR+NIR) Formula (2)

[0045] In formula (2), NIR indicates the reflectance of the near-infrared light image supplied from the second camera 22. The correction unit 14 calculates the reflectances of the near-infrared light image and the short-wave infrared light image using a known method, and calculates NDSI using formula (2) based on these reflectances.

[0046] NDSI is a value between -1 and +1. Utilizing the characteristic that in soil such as sand and asphalt, the reflectance of short-wave infrared light is large and the reflectance of near-infrared light is small, the correction unit 14 of the second embodiment determines a region within a predetermined threshold range (for example, a range near 0 such as -0.05 or more and 0.05 or less) of NDSI as a soil region (i.e., a non-recognition target region). In the example of FIG. 6, it is assumed that the overlapping region DU2 between the recognition frame DE3 and the shadow SH1 has an NDSI within the predetermined threshold range. In this case, the correction unit 14 determines that the overlapping region DU2 between the recognition frame DE3 and the shadow SH1 is a non-recognition target region with an NDSI within the predetermined threshold range, and corrects the recognition frame DE3 to exclude this overlapping region DU2 from the recognition frame DE3.

[0047] FIG. 7 is a diagram schematically showing an example of the recognition image F2 after the recognition frame is corrected. The correction unit 14 of the second embodiment performs a masking process so that the NDSI covers the overlapping region DU2 within a predetermined threshold range with the mask M2. The correction unit 14 generates a corrected recognition frame DE4 such that the boundary on the mask M2 side in the recognition frame DE3 is at a position in contact with the mask M2. For example, the correction unit 14 generates the corrected recognition frame DE4 by correcting the recognition frame DE3 so that the lower frame of the recognition frame is in contact with the upper end UP2 of the mask M2. Thereby, the range of the object is corrected.

[0048] FIG. 8 is a flowchart showing an example of the process S200 of the recognition processing method according to the second embodiment. Steps S201 to S203 and S207 in FIG. 8 are the same as steps S101 to S103 and S107 in FIG. 4, and thus the description thereof is omitted.

[0049] In step S204, the correction unit 14 calculates the NDSI for each position in the region within the recognition frame of the object to be recognized in the correction image.

[0050] In step S205, the correction unit 14 determines whether there is a region within the recognition frame of the object to be recognized where the NDSI is within the threshold range. If there is a region where the NDSI is within the threshold range (Yes in step S205), the process S200 proceeds to step S206. If there is no region where the NDSI is within the threshold range (No in step S205), the process S200 ends.

[0051] In step S206, the correction unit 14 masks the region within the recognition frame of the object to be recognized where the NDSI is within the threshold range. Thereafter, in step S207, the range of the object is corrected, and the process S200 ends.

[0052] In the second embodiment, the second acquisition unit 12 acquires, as a plurality of correction images, a first correction image generated by imaging light having a wavelength in the wavelength band of short-wave infrared light, and a second correction image generated by imaging near-infrared light having a wavelength in a wavelength band different from the wavelength band of short-wave infrared light. The correction unit 14 calculates an index indicating the degree of light reflection by soil based on the light reflectance in the range of each object in the first correction image and the second correction image. According to this configuration, false detection due to the shadow of an object to be recognized on the soil on the ground surface can be suppressed, so that the object to be recognized can be appropriately detected in the image recognition process. Summarizing the first embodiment and the second embodiment, the second acquisition unit 12 acquires, as a plurality of correction images, a first correction image generated by imaging light having a wavelength in the wavelength band of short-wave infrared light, and a second correction image generated by imaging light having a wavelength in a wavelength band different from the wavelength band of short-wave infrared light. The correction unit 14 calculates an index indicating the degree of light reflection by moisture or soil based on the light reflectance in the range of each object in the first correction image and the second correction image.

[0053] Hereinafter, some modification examples of the present disclosure will be described.

[0054] In the above embodiment, the first camera 21 is a far-infrared camera, but it is not limited thereto. For example, it may be a visible light camera that generates a visible light image.

[0055] In the above embodiment, the range of the object is corrected based on NDWI or NDSI, but it is not limited thereto. For example, the range of the object may be corrected based on the ratio of the reflectances of lights in different wavelength bands. For example, a region where the ratio of the reflectance of short-wave infrared light to the reflectance of near-infrared light is greater than a predetermined value may be determined as a water area and the range of the object may be corrected. Further, for example, the range of the object may be corrected based on the ratio of the intensities of lights in different wavelength bands. The ratio of the reflectances and the ratio of the intensities of the lights are examples of the degree of light reflection.

[0056] In the above-described embodiment, the correction unit 14 calculates NDWI or NDSI for each position of the region within the recognition frame. However, the present invention is not limited to this, and NDWI or NDSI may be calculated for the entire correction image, and the values within the recognition frame may be extracted and used.

[0057] In the above-described embodiment, the correction unit 14 calculates NDWI or NDSI. However, the present invention is not limited to this. For example, the correction unit 14 may calculate a normalized difference vegetation index (NDVI) or a normalized difference snow index (NDSI), etc., and identify vegetation regions, snow accumulation regions, etc. around the object to be recognized as non-recognition target regions.

[0058] FIG. 9 is a block diagram schematically showing the functional configuration of the recognition processing apparatus 10 according to a modification of the first embodiment. This modification is different from the first embodiment in that it further includes a rainfall detection unit 19 that detects the rainfall amount per unit time. The rainfall detection unit 19 detects the rainfall amount per unit time based on the detection result of raindrops by a raindrop sensor 23 installed on a multi-functional utility pole or the like. When the rainfall amount per unit time exceeds a predetermined rainfall threshold value, the rainfall detection unit 19 instructs the correction unit 14 not to execute the process of correcting the range of the above-described object. Thereby, it is possible to prevent the correction unit 14 that has received this instruction from executing the process of correcting the range of the object. Here, during rainy days, it is assumed that NDWI cannot be accurately calculated due to the influence of rainfall, the range of the object is erroneously corrected, and the object to be recognized is not appropriately detected. According to this modification, it is possible to suppress the range of the object from being erroneously corrected during rainy days.

[0059] As described above, the present disclosure has been described with reference to the above-described embodiments. However, the present disclosure is not limited to the above-described embodiments, and the present disclosure also includes combinations or substitutions of the respective configurations shown in the embodiments as appropriate.

Explanation of Reference Numerals

[0060] 10… Recognition processing device, 11… First acquisition unit, 12… Second acquisition unit, 13… Recognition unit, 14… Correction unit, 15… Output control unit, 16… Visible light image acquisition unit, 17… Short-wave infrared light image acquisition unit, 18… Near-infrared light image acquisition unit, 19… Rainfall detection unit, 21… First camera, 22… Second camera, 23… Raindrop sensor, 30… Output device.

Claims

1. A first acquisition unit that acquires a recognition image; A recognition unit that recognizes an object from the recognition image; A second acquisition unit that acquires a plurality of correction images generated by imaging light of different wavelengths, each including the range of the recognized object; A correction unit that corrects the range of the object based on the degree of light reflection in the range of the object in each of the plurality of correction images; A recognition processing apparatus comprising the above.

2. The second acquisition unit acquires, as the plurality of correction images, a first correction image generated by imaging light having a wavelength in the wavelength band of short-wave infrared light, and a second correction image generated by imaging light having a wavelength in a wavelength band different from the wavelength band of the short-wave infrared light, The correction unit calculates an index indicating the degree of light reflection by moisture or soil based on the light reflectance in the range of the object in each of the first correction image and the second correction image. The recognition processing apparatus according to claim 1.

3. The correction unit: Based on the degree of light reflection, identifies a non-recognition target area in the range of the object where there is a non-recognition target object different from the recognition target object; The recognition processing apparatus according to claim 1 or 2, wherein the range of the object is corrected such that the boundary on the non-recognition target area side in the range of the object is at a position in contact with the non-recognition target area.

4. A step of acquiring a recognition image; A step of recognizing an object from the recognition image; A step of acquiring a plurality of correction images generated by imaging light of different wavelengths, each including the range of the recognized object; A step of correcting the range of the object based on the degree of light reflection in the range of the object in each of the plurality of correction images; A recognition processing method comprising the above.

5. A recognition processing program that causes a computer to: Execute a step of acquiring a recognition image; Execute a step of recognizing an object from the recognition image; Execute a step of acquiring a plurality of correction images generated by imaging light of different wavelengths, each including the range of the recognized object; Execute a step of correcting the range of the object based on the degree of light reflection in the range of the object in each of the plurality of correction images. ​

Images