Image processing device, image processing method, and program

Abstract

To provide a technique which can enhance detection accuracy of a human body in a video of a camera.SOLUTION: An image processing device comprises: a video acquisition unit which acquires a video captured by a camera; a human body detection unit which detects a human body in the video acquired by the video acquisition unit; a dynamic body detection unit which detects a dynamic body in the video acquired by the video acquisition unit; a human body candidate specification unit which specifies an image of a region of a human body candidate from an image of the region detected with the human body detection by the human body detection unit on the basis of a coincidence degree between the image of the region detected with the human body detection by the human body detection unit and the image of the region detected with the dynamic body detection by the dynamic body detection unit; and a determination unit which determines whether the image of the region of the human body candidate is the image of the human body on the basis of a coincidence degree between the image of the region of the human body candidate specified by the human body candidate specification unit and an image for reference of an object erroneously detected as a human body.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a technique for detecting a human body in a camera image. [Background technology]

[0002] In surveillance using network cameras (IP cameras), there is a demand for improving the accuracy of detecting human bodies based on images captured by network cameras installed in buildings.

[0003] Therefore, a technology has been proposed that reduces false detection of human bodies by using differences (inter-frame differences, background differences, etc.) in images captured by a network camera. Patent Document 1 discloses a technology that detects human bodies from images by identifying the background and human bodies for a detected moving object using a dictionary. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2017-138922 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, in the conventional technology, since human bodies are detected based on moving object detection, it is not possible to detect stationary human bodies in a video. In addition, in the conventional technology, the background of an image in which a moving object is detected is compared with background images registered in a dictionary, so in order to improve the accuracy of human body detection, it is necessary to register a huge variety of background images in the dictionary.

[0006] The present invention has been made in consideration of the above circumstances, and provides a technique that can improve the accuracy of detecting a human body in a camera image. [Means for solving the problem]

[0007] In order to achieve the above object, the present invention employs the following configuration.

[0008] A first aspect of the present invention is an image processing device comprising: an image acquisition unit that acquires an image captured by a camera; a human body detection unit that performs human body detection in the image acquired by the image acquisition unit; a motion detection unit that performs motion detection in the image acquired by the image acquisition unit; a human body candidate identification unit that identifies an image of a human body candidate region from an image of a region detected by human body detection by the human body detection unit based on a degree of agreement between an image of a region detected by human body detection by the human body detection unit and an image of a region detected by motion detection by the motion detection unit; and a determination unit that determines whether or not the image of the human body candidate region is an image of a human body based on a degree of agreement between the image of the human body candidate region identified by the human body candidate identification unit and a reference image of a target that has been erroneously detected as a human body. This makes it possible to accurately detect a stationary human body as a human body and to suppress a phenomenon in which a target other than a human body is erroneously detected as a human body.

[0009] The human body candidate identifying unit may identify the human body candidate region based on a degree of agreement indicated by a distance between images calculated using coordinate information of an image of a region detected by human body detection by the human body detection unit and coordinate information of an image of a region detected by moving body detection by the moving body detection unit. This makes it possible to identify a region that is a human body candidate that is highly likely to be a human body from the camera image.

[0010] The determination unit may use coordinate information of the image of the human body candidate region and coordinate information of the reference image to determine whether the image of the human body candidate region is an image of a human body based on a degree of agreement indicated by a luminance difference between pixels constituting the image of the human body candidate region, excluding pixels of moving body parts, and corresponding pixels of the reference image. This makes it possible to specify an image of a human body with higher accuracy from the image of the human body candidate region detected in the camera video.

[0011] The determination unit may use, as the reference image, a first image that is not identified as an image of a region of a human body candidate by the human body candidate identification unit among images of a region detected by human body detection by the human body detection unit, or a second image that is not identified as an image of a human body by the determination unit among images of a human body candidate identified by the human body candidate identification unit. Furthermore, the determination unit may determine whether or not to use the first image as the reference image based on a luminance difference between the first image and the reference image already used, and may determine whether or not to use the second image as the reference image based on a luminance difference between the second image and the reference image already used. This allows for more accurate determination of erroneous detection of a human body using a reference image.

[0012] The present invention can also be understood as an image processing method including at least a part of the above-mentioned processing, a program for causing a computer to execute the method, or a computer-readable recording medium on which such a program is non-temporarily recorded. The above-mentioned configurations and processing can be combined with each other to constitute the present invention as long as no technical contradiction occurs.

[0013] The present invention can also be understood as an image processing method including at least a part of the above-mentioned processing, a program for causing a computer to execute the method, or a computer-readable recording medium on which such a program is non-temporarily recorded. The above-mentioned configurations and processing can be combined with each other to constitute the present invention as long as no technical contradiction occurs. Effect of the Invention

[0014] According to the present invention, it is possible to improve the accuracy of detecting a human body in a camera image. [Brief description of the drawings]

[0015] [Figure 1] FIG. 1 is a diagram illustrating a schematic configuration example of an image processing apparatus to which the present invention is applied. [Diagram 2]FIG. 2 is a block diagram illustrating an example of the configuration of an image processing device according to an embodiment. [Diagram 3] FIG. 3 is a flowchart showing an example of a processing flow of a PC according to an embodiment. [Figure 4] FIG. 3 is another flowchart showing an example of a process flow of the PC according to an embodiment. [Diagram 5] 5A to 5C are diagrams illustrating a specific example of image processing according to an embodiment. [Figure 6] 6A to 6C are other diagrams each showing a schematic example of image processing according to an embodiment. [Figure 7] 7A to 7C are diagrams illustrating an example of calculation of the degree of coincidence of images according to one embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] <Application Examples> An application example of the present invention will be described. In the conventional technology, a human body is detected based on motion detection using differences such as inter-frame differences and background differences in images captured by a network camera. However, in the conventional technology, since a human body is detected based on motion detection, a stationary human body in an image cannot be detected. In addition, in the conventional technology, since the background of an image in which a moving body is detected is compared with a background image registered in a dictionary, in order to improve the accuracy of human body detection, it is necessary to register a huge variety of background images in the dictionary.

[0017] FIG. 1 is a block diagram showing an example of the configuration of an image processing device 100 to which the present invention is applied. The image processing device 100 includes a video acquisition unit 101, a human body detection unit 102, a moving body detection unit 103, a human body candidate identification unit 104, and a determination unit 105. The video acquisition unit 101 acquires a video captured by a network camera 200 as an example of a fixed camera. The human body detection unit 102 performs human body detection in the acquired video. The moving body detection unit 103 performs moving body detection in the acquired video. The human body candidate identification unit 104 identifies an image of a human body candidate based on an image detected by the human body detection unit 102 and an image detected by the moving body detection unit 103. The determination unit 105 determines whether or not the image of the identified human body candidate is an image of a human body. More specifically, the determination unit 105 determines whether or not the image of the human body candidate is an image of a human body based on the degree of coincidence between the image of the identified human body candidate and an image of a target that may be erroneously detected as a human body.

[0018] According to the image processing device 100 of the present invention, it is possible to improve the accuracy of detecting a human body in an image captured by a camera.

[0019] <Description of the embodiment> An embodiment of the present invention will be described. Fig. 2 is a schematic diagram showing a rough configuration example of an image processing system according to this embodiment. The image processing system 1 according to this embodiment has a PC 100 (personal computer; image processing device), a network camera 200, and a display device 300. The PC 100 and the network camera 200 are connected to each other by wire or wirelessly, and the PC 100 and the display device 300 are connected to each other by wire or wirelessly.

[0020] In this embodiment, as an example, it is assumed that a network camera 200 installed in an outdoor building captures images of roads, houses, trees, and the like adjacent to the building. The network camera 200 outputs a video consisting of a plurality of frames of captured images to the PC 100. The PC 100 detects a moving object in the video captured by the network camera 200, determines a human body from the detected moving object, and outputs information related to the determined human body to a display device. Examples of the display device include a display and an information processing terminal (such as a smartphone).

[0021] In this embodiment, the PC 100 is a separate device from the network camera 200 and the display device 300, but the PC 100 may be configured integrally with the network camera 200 or the display device 300. The location where the PC 100 is installed is not particularly limited. For example, the PC 100 may be installed in the same location as the network camera 200. The PC 100 may also be a computer on the cloud.

[0022] PC 100 has an input unit 110, a control unit 120, a storage unit 130, and an output unit 140. Control unit 120 has a human body candidate identifying unit 121 and a determination unit 122. Furthermore, human body candidate identifying unit 121 has a human body detection unit 123, a moving body detection unit 124, and a detection region comparison unit 125. Furthermore, determination unit 122 has a non-moving pixel extraction unit 126, a false detection list determination unit 127, and a false detection list update unit 128.

[0023] The input unit 110 corresponds to the image acquisition unit of the present invention and receives images captured by the network camera 200. The captured image is acquired from the network camera 200 and output to the control unit 120. Note that the network camera 200 does not have to be an optical camera, and may be a thermal camera or the like.

[0024] The control unit 120 includes a central processing unit (CPU), a random access memory (RAM), a read only memory (ROM), and the like, and controls each unit in the PC 100 and performs various information processing.

[0025] The human body detection unit 123 detects a human body from an image within the angle of view of the network camera 200, and detects the target as a rectangular area. The moving object detection unit 124 detects a moving object from the image, and detects the target as a rectangular area. The detection area comparison unit 125 compares an image of the rectangular area detected by the human body detection unit 123 with an image of the rectangular area detected by the moving object detection unit 124, calculates the degree of agreement, and identifies a rectangular area of ​​a human body candidate based on the calculated degree of agreement.

[0026] The non-moving body pixel extraction unit 126 extracts pixels of non-moving bodies from the image of the rectangular region of the human body candidate, excluding pixels detected as moving bodies. The false detection list comparison unit 127 compares the image from which the non-moving body pixels are extracted by the non-moving body pixel extraction unit 126 with an image that is falsely detected as a human body. In this embodiment, the image that is falsely detected as a human body is stored in the storage unit 130 as a reference image for the false detection list. The false detection list update unit 128 updates the reference image for the false detection list stored in the storage unit 130 with an image of a rectangular region that is determined not to be a human body among rectangular regions detected by human body detection by the human body detection unit 123.

[0027] The storage unit 130 stores reference images for the false detection list, as well as programs executed by the control unit 120 and various data used in the processing executed by the control unit 120. For example, the storage unit 130 is an auxiliary storage device such as a hard disk drive or a solid state drive. The output unit 240 outputs information notifying the result of the human body determination by the processing of the control unit 120 to the display device 300. Note that information related to the result of the human body determination by the control unit 120 may be stored in the storage unit 130 and output from the output unit 140 to the display device 300 at any timing.

[0028] 3 and 4 are flowcharts showing an example of a process flow of the PC 100. The process flow shown in Fig. 4 is a process flow showing a subroutine of step S309 in Fig. 3. As an example, the PC 100 starts the process flow of Fig. 3 and 4 after the power is turned on.

[0029] First, the storage of a reference image of the false detection list used in the process flow shown in Figs. 3 and 4 in the storage unit 130 will be described. Here, it is assumed that a reference image of the false detection list is not stored in the storage unit 130. However, an image of a rectangular area of ​​an object that has been falsely detected in a video captured by the network camera 200 and coordinate information that can specify the range (position) of the image may be stored in advance in the storage unit 130 as a reference image and its coordinate information. Here, examples of the coordinate information include the coordinates of the upper left corner and the coordinates of the lower right corner of the rectangular area, and the center coordinate of the rectangular area. Also, instead of or in addition to an object that has been falsely detected in the past, an image of an object that may be falsely detected and its coordinate information may be stored in the storage unit 130.

[0030] In step S301, the input unit 110 of the PC 100 acquires an image of the network camera 200 connected to the PC 100. The image acquired by the input unit 110 is transmitted to the control unit 120. Note that the image acquired by the input unit 110 may be stored in the storage unit 130, and the control unit 120 may acquire the image stored in the storage unit 130 and execute the following processing.

[0031] Next, in step S302, the human body detection unit 123 of the control unit 120 performs a human body detection process on the video acquired in step S301, and detects the detected object as a rectangular area in the image constituting the video. The human body detection unit 123 also acquires coordinate information of each detected rectangular area. Next, in step S303, the moving object detection unit 124 of the control unit 120 performs a moving object detection process on the video acquired in step S301, and detects the detected object as a rectangular area in the image constituting the video. The moving object detection unit 124 also acquires coordinate information of each detected rectangular area.

[0032] 5A to 5C show an example of the processing results in steps S301 to S303. FIG. 5A shows an example of an image constituting the video acquired by the input unit 110 from the network camera 200 in step S301. Here, the video of a house 401, a sign 402, a road 403, a tree 404, and a walking person 405 are acquired by the network camera 200. FIG. 5B shows an example in which the human body detection unit 123 detects an object detected by the human body detection process as a rectangular area in step S302. As shown in FIG. 5B, the tree 404 and the person 405 are detected as rectangular areas 406 and 407 in the video, respectively. FIG. 5C shows an example in which the moving object detection unit 124 detects a moving object detected by the moving object detection process as a rectangular area in step S303. As shown in FIG. 5C, the walking person 405 is detected as a rectangular area 408 in the video.

[0033] 3, when the moving object detection unit 120 completes the process of step S303, the control unit 120 repeatedly executes the processes of steps S304 to S311 for each rectangular area detected in step S302. Therefore, the loop process of steps S304 to S311 is executed a number of times corresponding to the number of rectangular areas detected in step S302.

[0034] In step S304, the detection area comparison unit 125 compares the rectangular area currently being processed in the loop with the rectangular area of ​​the moving object detected in step S303, and calculates the degree of match between the two rectangular areas. Specifically, the detection area comparison unit 125 calculates the degree of match based on the IoU (Intersection over Union), inclusion rate, distance, etc. of the two rectangular areas. If the calculated degree of agreement is equal to or greater than a predetermined threshold (S304: YES), the region comparison unit 125 proceeds to step S305, since the object in the rectangular region detected by the human body detection in step S302 is also detected as a moving object. On the other hand, if the calculated degree of agreement is less than the threshold (S304: NO), the detection region comparison unit 125 proceeds to step S309, since the object in the rectangular region detected by the human body detection in step S302 is not detected as a moving object. The image processed from step S304 to step S309 corresponds to the first image that was not identified as an image of a region of a human body candidate by the human body candidate identification unit in the present invention.

[0035] In the example shown in FIGS. 5A to 5C, the rectangular area 406 of the tree 404 detected in step S302 has a low degree of coincidence with the rectangular area 408 of the person 405 detected in step S303, so the control unit 120 advances the process from step S304 to step S309. Next, in step S401, the control unit 120 determines whether or not a reference image of the erroneous detection list is stored in the storage unit 130. Here, since a reference image is not stored in the storage unit 130 (S401: NO), the control unit 120 advances the process to step S405, stores the image of the rectangular area 406 and the coordinate information of the image in the storage unit 130 as a reference image and its coordinate information, ends the subroutine in FIG. 4, and returns to the process flow in FIG. 3.

[0036] In this way, if the reference image for the erroneous detection list is not stored in the storage unit 130, An image in which an object that is not a human body is detected as a human body in the video captured by network camera 200 (an image of rectangular area 406 in the example of FIG. 5B) and coordinate information are stored in storage unit 130.

[0037] Next, the process flow when a reference image of the erroneous detection list is stored in storage unit 130 will be described with reference to the process flows in Figs. 3 and 4 along with another video example. Note that since the processes in steps S301 to S304 are the same as those described above, the details of the other processes will be described below. Also, it is assumed that storage unit 130 stores an image of rectangular area 406 in Fig. 5B and its coordinate information as a reference image. Also, it is assumed that the number of pixels in this reference image is 50 pixels.

[0038] 6A to 6C show an example of the processing results in steps S301 to S303. FIG. 6A shows an example of an image constituting the video acquired by the input unit 110 from the network camera 200 in step S301. Here, the video of a house 401, a sign 402, a road 403, a tree 404, walking people 409 and 410, and a car 411 running on the road 403 are acquired by the network camera 200. FIG. 6B shows an example of a rectangular area detected by the human body detection unit 123 through the human body detection process in step S302. As shown in FIG. 6B, the tree 404 and people 409 and 410 are detected as rectangular areas 412, 413, and 414, respectively, in the video. FIG. 6C shows an example of a rectangular area detected by the moving object detection unit 124 through the moving object detection process in step S303. As shown in FIG. 6C, walking people 409, 410 and a moving car 411 are detected as rectangular regions 415, 416, and 417, respectively, in the video.

[0039] Returning to FIG. 3, in step S305, the false detection list determination unit 127 determines whether the distance between the rectangular area of ​​the human body candidate that is the target of the current loop processing and the reference image of the false detection list stored in the storage unit 130 is less than a predetermined threshold. Specifically, the false detection list determination unit 127 determines whether the distance (e.g., center distance) calculated using the coordinate information of the rectangular area of ​​the human body candidate that is the target of the current loop processing and the coordinate information of the reference image is less than or equal to a threshold. If the calculated distance is less than or equal to the threshold (S305: YES), there is a possibility that the object in the rectangular area of ​​the human body candidate that is the target of the current loop processing is the object of the reference image of the false detection list, so the control unit 120 advances the process to step S306. On the other hand, if the calculated distance is longer than the threshold (S305: NO), the object in the rectangular area of ​​the human body candidate that is the target of the current loop processing is different from the object of the reference image of the false detection list, and is not considered to have been erroneously detected, and the control unit 120 advances the process to step S308.

[0040] In the example shown in FIG. 6A to FIG. 6C, the image of rectangular region 412 of tree 404 detected by human body detection and the image of rectangular region 406 stored in storage unit 130 are images of rectangular regions detected for the same object, tree 404, so the distance between the images is equal to or less than the threshold. Therefore, the process for the image of rectangular region 412 of the human body candidate proceeds from step S305 to step S306. On the other hand, the images of rectangular regions 413 and 414 of persons 409 and 410 detected by human body detection and the image of rectangular region 406 stored in storage unit 130 are images of rectangular regions detected at different positions, so the distance between the images is longer than the threshold. Therefore, the process for the images of rectangular regions 413 and 414 of the human body candidate proceeds from step S305 to step S308.

[0041] Returning to Fig. 3, in step S306, non-moving body pixel extraction unit 126 extracts pixels excluding pixels corresponding to moving body parts from among the pixels constituting the image of the rectangular region of the human body candidate that is the target of the current loop processing, to generate an image. Then, control unit 120 advances the process to step S307. In step S307, false detection list determination unit 127 calculates the degree of match between the image generated in step S306 and the reference image of the false detection list stored in storage unit 130, and determines whether the calculated degree of match is equal to or lower than a predetermined threshold. If the calculated degree of match is equal to or less than the threshold (S307: YES), the object within the rectangular region of the human body candidate that is the object of the current loop processing is deemed not to be an object of the reference image in the false detection list, and control unit 120 advances the process to step S308. On the other hand, if the calculated degree of match is greater than the threshold (S307: NO), the object within the rectangular region of the human body candidate that is the object of the current loop processing may be an object of the reference image in the false detection list, so control unit 120 advances the process to step S309. Note that the image that is processed from step S307 to step S309 corresponds to the second image that is not determined to be an image of a human body by the determination unit in the present invention.

[0042] An example of the processing in steps S306 and S307 will now be described with reference to Figures 7A to 7C. As an example, the threshold in step S307 is set to 0.3. Figure 7A shows a rectangular area 412 detected by human body detection in step S302. The image of rectangular area 412 is made up of pixels 410 of trees 404, pixels 418 of a running car 411, and pixels 419 of other parts.

[0043] In step S306, the non-moving object pixel extraction unit 126 generates an image (hereinafter also referred to as a "non-moving object pixel image") composed of pixels excluding pixel 418 corresponding to the car 411, which is a moving object, from among the pixels constituting the image of the rectangular area 412. Here, as an example, it is assumed that the total number of pixels in the image of the rectangular area 412 is 50 pixels, the number of pixels in the pixel 418 is 20 pixels, and the remaining number of pixels 410 and 419 is 30 pixels.

[0044] In step S307, the misdetection list comparison unit 127 calculates the degree of match using the number of pixels of the image generated in step S306 and the number of pixels of the reference image of the misdetection list stored in the memory unit 130, using the following formula (1).

number

[0045] In addition, "the number of pixels in the non-moving object pixel image whose brightness difference with the reference image is below a threshold value" refers to, in the example of Figure 7A, the reference image is, for example, an image of rectangular area 406 of tree 404 stored in memory unit 130, and the non-moving object pixel image is an image obtained by excluding pixel 418 from the entire image of rectangular area 412, and refers to the number of pixels in these two images whose brightness difference between pixels at corresponding coordinates is below a threshold value based on their respective coordinate information.

[0046] In the example of FIG. 7A, the "number of pixels in the reference image" is 50 pixels, and the "number of pixels in the non-moving pixel image" is 30 pixels. Also, the "number of pixels in the non-moving pixel image whose brightness difference with the reference image is below the threshold" is 30 pixels, because pixels 410 and 419 of the image in rectangular area 412 that is the target of the current loop processing have no brightness difference with the pixels in the image in rectangular area 406 of the reference image and are below the threshold. Therefore, in this case, the degree of match is calculated as 30 / 80 = 0.375 according to formula (1). As a result, for the image of rectangular area 412, the process advances from step S307 to step S309.

[0047] Another example of a non-moving object pixel image is shown in Fig. 7B and Fig. 7C. In the example shown in Fig. 7B, a person 421 walking in front of a tree 404 in the video of the network camera 200 is detected as a rectangular area 422 by the human body detection in step S302. Here, the number of pixels in the entire image of the rectangular area 422 is 10 pixels, and the number of pixels in the image of the person 421 is 5 pixels. In this case, in step S307, as shown in Fig. 7C, an image in which pixel 424 of the person 421 is excluded as a moving object pixel from among the pixels in the entire image of the rectangular area 422 is generated as a non-moving object pixel image. Therefore, this non-moving object pixel image is composed of the tree 404, pixels 423 other than the person 421, and pixels 425 of the tree 404.

[0048] In the example of FIG. 7B and FIG. 7C, the "number of pixels in the reference image" is 50 pixels, and the "number of pixels in the non-moving object pixel image" is 5 pixels. Also, the "number of pixels in the non-moving object pixel image whose luminance difference from the reference image is equal to or less than the threshold" is 5 pixels because pixels 423 and 425 of the image in rectangular area 422 currently being processed in the loop have no luminance difference from the pixels in the image in rectangular area 406 of the reference image and are equal to or less than the threshold. Therefore, in this case, the degree of match is calculated as 5 / 55=0.091 according to formula (1), which is equal to or less than the threshold of 0.3. As a result, for the image in rectangular area 422, the process proceeds from step S307 to step S308.

[0049] 3, in step S308, the erroneous detection list determination unit 127 determines that the object in the rectangular region of the human body candidate that is the target of the current loop process is a human body. Then, the control unit 120 repeatedly executes the above-mentioned loop process for the remaining rectangular regions detected in step S302.

[0050] Next, the processing of the subroutine executed in step S309 will be described with reference to Fig. 4. In step S401, control unit 120 determines whether or not a reference image for the erroneous detection list is stored in storage unit 130. If a reference image for the erroneous detection list is stored in storage unit 130 (S401: YES), control unit 120 advances the processing to step S402. If a reference image for the erroneous detection list is not stored in storage unit 130 (S401: NO), control unit 120 advances the processing to step S405.

[0051] In step S402, the erroneous detection list determination unit 127 calculates the distance between the image of the rectangular region currently being processed in the loop and the reference image in the erroneous detection list, and determines whether the calculated distance is equal to or greater than a threshold. If the calculated distance is equal to or greater than the threshold (S402: YES), the control unit 120 assumes that the image of the rectangular region currently being processed in the loop is an erroneously detected image that is different from the reference image in the erroneous detection list, and proceeds to step S405. On the other hand, if the calculated distance is less than the threshold (S402: NO), the control unit 120 proceeds to step S403.

[0052] In step S403, the false detection list determination unit 127 calculates the image size ratio between the image of the rectangular region currently being processed in the loop and the reference image in the false detection list, and determines whether the calculated ratio is equal to or greater than a threshold. If the calculated image size ratio is equal to or greater than the threshold (S403: YES), the control unit 120 determines that the image is an object other than the reference image in the false detection list that has been falsely detected, and proceeds to step S405. On the other hand, if the calculated size ratio is less than the threshold (S403: NO), the control unit 120 proceeds to step S404.

[0053] In step S404, the erroneous detection list determination unit 127 determines whether the Using the coordinate information of the image of the rectangular region and the reference image of the false detection list, the ratio of the number of pixels whose brightness difference with the corresponding pixel of the reference image of the false detection list is equal to or greater than a threshold is calculated for all pixels in the image of the rectangular region currently being processed in the loop. Then, the false detection list determination unit 127 determines whether the calculated ratio is equal to or greater than a threshold. If the calculated ratio is equal to or greater than the threshold (S404: YES), the control unit 120 advances the process to step S405 to replace the reference image of the false detection list with the image of the rectangular region currently being processed in the loop. On the other hand, if the calculated ratio is less than the threshold (S404: NO), the control unit 120 ends the process of this subroutine.

[0054] In step S405, if the process has proceeded from step S402 or step S403 to step S405, the false detection list update unit 128 stores the image of the rectangular area that is the current target of the loop processing in the storage unit 130 as a new reference image for the false detection list. In addition, if the process has proceeded from step S404 to step S405, the false detection list update unit 128 replaces the reference image in the false detection list with the image of the rectangular area that is the current target of the loop processing.

[0055] 3, among the images of the rectangular region detected by the human body detection in step S302, an image different from the reference image in the false detection list already stored in the storage unit 130 is stored as a new reference image in the storage unit 130. In addition, the luminance of the object in the video changes over time, for example, due to the weather or the AE (Automatic Exposure) function of the network camera 200. For this reason, even if an image among the images of the rectangular region detected by the human body detection in step S302 is the same as the reference image in the false detection list already stored in the storage unit 130, by replacing the image based on the luminance determination in step S404, it is expected that the reference image in the false detection list can be updated according to the change in luminance of the object in the video captured by the network camera 200, thereby improving the accuracy of false detection.

[0056] As described above, according to the image processing device of this embodiment, a human body can be detected even if the human body is stationary in the image captured by the camera, and even if an object other than a human body is detected as a moving object due to the presence of a moving object moving around the object, the object is deemed to have been falsely detected based on the degree of similarity with the images in the false detection list, thereby enabling more accurate human body detection.

[0057] <Other> The above embodiment merely exemplifies the configuration of the present invention. The present invention is not limited to the above specific form, and various modifications are possible within the scope of the technical idea. Modifications of the above embodiment are described below. In the following description, the same reference numerals are used for configurations similar to those of the above embodiment, and detailed description is omitted. In addition, the configurations and processes of the above embodiment and each of the modifications described below may be appropriately combined with each other.

[0058] <Appendix 1> An image acquisition unit (110) that acquires an image captured by a camera; a human body detection unit (123) that detects a human body in the image acquired by the image acquisition unit; a moving object detection unit (124) that detects a moving object in the image acquired by the image acquisition unit; a human body candidate identifying unit (1) that identifies an image of a region of a human body candidate from the image of the region detected by human body detection by the human body detection unit based on a degree of coincidence between an image of the region detected by human body detection by the human body detection unit and an image of the region detected by moving body detection by the moving body detection unit; 21) and a determination unit (122) that determines whether or not the image of the region of the human body candidate identified by the human body candidate identification unit is an image of a human body based on a degree of coincidence between an image of the region of the human body candidate identified by the human body candidate identification unit and a reference image of a target that has been erroneously detected as a human body; 13. An image processing device comprising:

[0059] <Appendix 2> A step of acquiring an image captured by a camera (S301); A step of detecting a human body in the acquired image (S302); A step (S303) of detecting a moving object in the acquired image; A step (S304) of identifying an image of a region of a human body candidate from the image of the region detected by the human body detection based on a degree of coincidence between the image of the region detected by the human body detection and the image of the region detected by the moving object detection; determining whether the image of the region of the identified human body candidate is an image of a human body based on a degree of coincidence between the image of the region of the identified human body candidate and a reference image of the object erroneously detected as a human body (S307, S308); 13. An image processing method comprising: [Explanation of symbols]

[0060] 100: Image processing device 110: Input unit 120: Control unit 122: Determination unit 123: Human body detection unit 124: Moving body detection unit 200 Network Cameras

Claims

1. an image acquisition unit that acquires an image captured by a camera; a human body detection unit that detects a human body in the image acquired by the image acquisition unit; a moving object detection unit that detects a moving object in the video acquired by the video acquisition unit; a human body candidate identifying unit that identifies an image of a region of a human body candidate from the image of the region detected by human body detection by the human body detection unit based on a degree of coincidence between an image of the region detected by human body detection by the human body detection unit and an image of the region detected by moving body detection by the moving body detection unit; a determination unit that determines whether or not the image of the region of the human body candidate identified by the human body candidate identification unit is an image of a human body based on a degree of coincidence between an image of the region of the human body candidate identified by the human body candidate identification unit and a reference image of a target that has been erroneously detected as a human body; It has the determination unit uses coordinate information of the image of the human body candidate region and coordinate information of the reference image to determine whether the image of the human body candidate region is an image of a human body based on a degree of agreement indicated by a luminance difference between pixels constituting the image of the human body candidate region, excluding pixels of moving body parts, and corresponding pixels of the reference image. Image processing device.

2. The image processing device according to claim 1, characterized in that the human body candidate identification unit identifies the region of the human body candidate based on a degree of similarity indicated by a distance between images calculated using coordinate information of an image of a region detected by human body detection by the human body detection unit and coordinate information of an image of a region detected by motion detection by the motion detection unit.

3. An image acquisition unit that acquires an image captured by a camera; a human body detection unit that detects a human body in the image acquired by the image acquisition unit; a moving object detection unit that detects a moving object in the video acquired by the video acquisition unit; a human body candidate identifying unit that identifies an image of a region of a human body candidate from the image of the region detected by human body detection by the human body detection unit based on a degree of coincidence between an image of the region detected by human body detection by the human body detection unit and an image of the region detected by moving body detection by the moving body detection unit; a determination unit that determines whether or not the image of the region of the human body candidate identified by the human body candidate identification unit is an image of a human body based on a degree of coincidence between an image of the region of the human body candidate identified by the human body candidate identification unit and a reference image of a target that has been erroneously detected as a human body; It has The determination unit is The present invention is characterized in that it determines whether or not to use the first image as the reference image based on a brightness difference between a first image, which is not identified as an image of a region of a human body candidate by the human body candidate identification unit among images of a region detected by human body detection by the human body detection unit, and the reference image that is already being used, or it determines whether or not to use the second image as the reference image based on a brightness difference between a second image, which is not determined as an image of a human body by the determination unit among images of human body candidates identified by the human body candidate identification unit, and the reference image that is already being used. Image processing device.

4. acquiring an image captured by a camera; performing human body detection in the captured image; performing moving object detection in the acquired video; identifying an image of a region of a human body candidate from the image of the region detected by the human body detection based on a degree of coincidence between the image of the region detected by the human body detection and the image of the region detected by the motion detection; using coordinate information of an image of the identified region of the human body candidate and coordinate information of a reference image of a target erroneously detected as a human body, determining whether the image of the region of the human body candidate is an image of a human body based on a degree of agreement indicated by a brightness difference between pixels constituting the image of the region of the human body candidate, excluding pixels of moving body portions, and corresponding pixels of the reference image; 13. An image processing method comprising:

5. A step of acquiring an image captured by a camera; performing human body detection in the captured image; performing moving object detection in the acquired video; identifying an image of a region of a human body candidate from the image of the region detected by the human body detection based on a degree of coincidence between the image of the region detected by the human body detection and the image of the region detected by the motion detection; determining whether the image of the region of the identified human body candidate is an image of a human body based on a degree of coincidence between an image of the region of the identified human body candidate and a reference image of a target erroneously detected as a human body; a step of determining whether or not to use the first image as the reference image based on a brightness difference between a first image, which is not identified as an image of a human body candidate region among images of the region detected by the human body detection, and the reference image that is already being used, or determining whether or not to use the second image as the reference image based on a brightness difference between a second image, which is not identified as an image of a human body among images that are identified as the region of the human body candidate, and the reference image that is already being used; 13. An image processing method comprising:

6. 6. A program for causing a computer to execute each step of the image processing method according to claim 4.

Images