JP7472772B2 - Image processing device, overload detection system, image processing method, and image processing program - Google Patents

Description

This invention relates to an image processing device, an overload detection system, an image processing method, and an image processing program, and in particular to an image processing device that identifies tire portions corresponding to the vehicle's tires in an image obtained by capturing an image of the vehicle, an overload detection system that includes the image processing device, an image processing method that is executed by the image processing device, and an image processing program that causes a computer to execute the image processing method.

Technology is known that determines whether a vehicle is overloaded from the amount of deformation of the tires based on the shape of the tires of a passing vehicle. For example, JP 2002-228423 A discloses a tire detection method that uses an imaging device to capture an image of the area around the tires of a passing vehicle from diagonally above and to the side, detects the edges of the image obtained to detect tire contour information, defines a pseudo-Hough transform process that detects ellipses for the detected tire contour information, and performs voting processing in the parameter space of the Hough transform at the edge points of the edge detection image to detect the tires.

However, the parts extracted as edges from the image captured by the imaging device are not limited to the outline of the tire, and may include parts other than the tire, such as the outline of the body or other parts. This can lead to the vehicle's body or other parts being mistakenly extracted as tires.

JP 2002-228423 A

This invention has been made to solve the above-mentioned problems, and one of the objects of this invention is to provide an image processing device that can accurately extract the tire portion corresponding to the vehicle's tires from an image obtained by capturing an image of the vehicle.

Another object of the present invention is to provide an overload detection system that can detect whether a vehicle is overloaded from an image obtained by capturing an image of the vehicle.

A further object of the present invention is to provide an image processing method capable of accurately extracting tire portions corresponding to the vehicle's tires from an image obtained by imaging the vehicle.

Yet another object of the present invention is to provide an image processing program capable of accurately extracting tire portions corresponding to the vehicle's tires from an image obtained by photographing the vehicle.

In order to achieve the above-mentioned object, according to one aspect of the present invention, an image processing device includes an acquisition means for acquiring a captured image obtained by photographing the side of a vehicle, a first processing means for performing a first image processing on the captured image and extracting edges contained in the image after the first image processing, a second processing means for performing a second image processing different from the first image processing on the captured image and extracting edges contained in the image after the second image processing, a first extraction means for extracting a first circle or ellipse by applying a Hough transform to the edges extracted by the first processing means, a second extraction means for extracting a second circle or ellipse by applying a Hough transform to the edges extracted by the second processing means, and an identification means for identifying a tire portion in the captured image that corresponds to the vehicle's tires based on the first circle or ellipse and the second circle or ellipse.

According to this aspect, an image obtained by photographing the side of a vehicle is obtained, and a first circle or ellipse is extracted by Hough transforming edges extracted from an image obtained by performing a first image processing on the image obtained by performing a second image processing on the image obtained by performing a Hough transform on the image obtained by performing a second image processing on the image obtained by performing a second circle or ellipse. Then, a tire portion corresponding to the tire of the vehicle in the image obtained by performing a first image processing on the image obtained by performing a second ... This makes it possible to provide an image processing device that can accurately extract tire portions corresponding to the vehicle's tires from an image obtained by capturing an image of the vehicle.

Preferably, each of the first image processing and the second image processing includes a smoothing process for smoothing an image, a differential image generating process for generating a differential image by differentiating the smoothed image, and a binarization process for binarizing the differential image, and the degree of smoothing in the smoothing process included in the first image processing is greater than the degree of smoothing in the smoothing process included in the second image processing, and the binarization threshold in the binarization process included in the first image processing is smaller than the binarization threshold in the binarization process included in the second image processing.

According to this aspect, since the degree of smoothing in the first image processing is greater than the degree of smoothing in the second image processing, the differential image generated by the first image processing has less noise than the differential image generated by the second image processing. Since the binarization threshold in the first image processing is smaller than the binarization threshold in the second image processing, edges with small changes in brightness in the captured image can be extracted from the differential image generated by the first image processing. At this time, since the differential image generated by the first image processing has less noise than the differential image generated by the second image processing, it is possible to prevent the noise from being erroneously detected as an edge. Furthermore, the differential image generated by the second image processing has more noise than the differential image generated by the first image processing, but since the binarization threshold in the second image processing is greater than the binarization threshold in the first image processing, it is possible to prevent the noise from being erroneously detected as an edge, and only edges with large changes in brightness in the captured image can be extracted from the differential image generated by the second image processing.

Preferably, each of the first image processing and the second image processing includes a reduction processing for reducing an image, a differential image generating processing for generating a differential image by differentiating the reduced image, and a binarization processing for binarizing the differential image, and the degree of reduction in the reduction processing included in the first image processing is greater than the degree of reduction in the reduction processing included in the second image processing, and the binarization threshold in the binarization processing included in the first image processing is smaller than the binarization threshold in the binarization processing included in the second image processing.

According to this aspect, since the degree of reduction in the first image processing is greater than the degree of reduction in the second image processing, the differential image generated by the first image processing has less noise than the differential image generated by the second image processing. Since the binarization threshold in the first image processing is smaller than the binarization threshold in the second image processing, edges with small changes in brightness in the captured image can be extracted from the differential image generated by the first image processing. At this time, since the differential image generated by the first image processing has less noise than the differential image generated by the second image processing, it is possible to prevent the noise from being erroneously detected as an edge. Furthermore, the differential image generated by the second image processing has more noise than the differential image generated by the first image processing, but since the binarization threshold in the second image processing is greater than the binarization threshold in the first image processing, it is possible to prevent the noise from being erroneously detected as an edge, and only edges with large changes in brightness in the captured image can be extracted from the differential image generated by the second image processing.

Preferably, the first extraction means is set with first diameter information determined based on the size of a tire mounted on the vehicle, and the second extraction means is set with second diameter information determined based on the size of a rim mounted on the vehicle.

According to this aspect, the first diameter information is determined based on the size of the tire mounted on the vehicle, so that the portion corresponding to the outer edge of the tire can be extracted as a first circle or ellipse, and the second diameter information is determined based on the size of the rim mounted on the vehicle, so that the portion corresponding to the outer edge of the rim can be extracted as a second circle or ellipse.

Preferably, the first extraction means sets a first threshold determined by the first diameter information as a threshold to be compared with the voting result in the Hough transform, and the first extraction means sets a second threshold determined by the second diameter information as a threshold to be compared with the voting result in the Hough transform.

According to this aspect, a first threshold determined by the first diameter information is set as a threshold to be compared with the voting result in the Hough transform, and a second threshold determined by the second diameter information is set as a threshold to be compared with the voting result in the Hough transform. Therefore, a first circle or ellipse corresponding to the diameter of the outer edge of the tire and a second circle or ellipse corresponding to the diameter of the outer edge of the rim can be extracted from the captured image.

Preferably, the vehicle further includes a determination means for determining whether the vehicle is overloaded based on the tire portion identified by the identification means.

In this way, it is easy to determine whether a vehicle is overloaded from captured images.

According to another aspect of the invention, an overload detection system includes the image processing device described above and an imaging means for imaging a vehicle and outputting the image obtained to the image processing device.

According to this aspect, it is possible to provide an overload detection system that can detect whether a vehicle is overloaded from an image obtained by capturing an image of the vehicle.

According to yet another aspect of the invention, the image processing method causes an image processing device to execute the following steps: an acquisition step of acquiring a captured image obtained by photographing the side of a vehicle; a first processing step of performing a first image processing on the captured image and extracting edges included in the image after the first image processing; a second processing step of performing a second image processing different from the first image processing on the captured image and extracting edges included in the image after the second image processing; a first extraction step of extracting a first circle or ellipse by applying a Hough transform to the edges extracted in the first processing step; a second extraction step of extracting a second circle or ellipse by applying a Hough transform to the edges extracted in the second processing step; and an identification step of identifying a tire portion in the captured image that corresponds to the vehicle's tire based on the first circle or ellipse and the second circle or ellipse.

According to this aspect, it is possible to provide an image processing method that can accurately extract tire portions corresponding to the vehicle's tires from an image obtained by capturing an image of the vehicle.

According to yet another aspect of the invention, the image processing program causes a computer to execute an acquisition step of acquiring a captured image obtained by photographing the side of a vehicle, a first processing step of performing a first image processing on the captured image and extracting edges included in the image after the first image processing, a second processing step of performing a second image processing different from the first image processing on the captured image and extracting edges included in the image after the second image processing, a first extraction step of extracting a first circle or ellipse by applying a Hough transform to the edges extracted in the first processing step, a second extraction step of extracting a second circle or ellipse by applying a Hough transform to the edges extracted in the second processing step, and an identification step of identifying a tire portion in the captured image that corresponds to the vehicle's tire based on the first circle or ellipse and the second circle or ellipse.

In accordance with this aspect, it is possible to provide an image processing program that can accurately extract tire portions corresponding to the vehicle's tires from an image obtained by capturing an image of the vehicle.

1 is a diagram showing an example of an overall overview of an overload detection system according to an embodiment of the present invention; FIG. 2 is a block diagram showing an example of a hardware configuration of the imaging apparatus. FIG. 2 is a block diagram showing an example of a hardware configuration of the image processing device. 2 is a diagram illustrating an example of functions of a CPU included in the image processing device according to the present embodiment. FIG. 11A and 11B are diagrams illustrating differences in parameters used in the first image processing and the second image processing. FIG. 11 is a diagram illustrating an example of a first extraction condition and a second extraction condition. 10 is a diagram showing an example of a relationship between a tire portion included in a captured image and a circle or an ellipse extracted by a Hough transform; 10A and 10B are diagrams illustrating a tire deformation determination amount. 13 is a flowchart showing an example of the flow of an overload detection process. 10 is a flowchart showing an example of the flow of a tire portion extraction process. 13 is a flowchart showing an example of the flow of an edge extraction process.

The following describes an image processing device according to an embodiment of the present invention, and an overload detection system equipped with the image processing device, with reference to the drawings. In the following description, identical parts are given the same reference numerals. Their names and functions are also the same. Therefore, detailed descriptions of them will not be repeated.

Figure 1 is a diagram showing an example of an overall overview of an overload detection system in this embodiment. The overload detection system 1 includes an imaging device 10 and an image processing device 20. The imaging device 10 is provided in a fixed position and in a fixed orientation beside an entry/exit lane of a motorway such as an expressway, for example, but is not limited to, and captures images of vehicles entering or exiting the motorway. The imaging device 10 includes a camera 11 (see Figure 2). The camera 11 outputs a captured image obtained by capturing an image of a vehicle traveling on a roadway RW.

The image processing device 20 acquires the captured image from the imaging device 10, processes the image, and determines whether the vehicle is overloaded based on the deformation of the vehicle's tires. The image processing device 20 determines whether the weight of the vehicle, including the weight of the load, exceeds a predetermined standard corresponding to the vehicle.

Note that while Figure 1 shows a case where only one side of the vehicle is photographed, both sides of the vehicle may be photographed to improve measurement accuracy.

Fig. 2 is a block diagram showing an example of a hardware configuration of an imaging device. Referring to Fig. 2, the imaging device 10 includes a camera 11, a control unit 12, a storage unit 13, and a communication unit 14. The control unit 12 is a central processing unit (CPU) that controls the entire imaging device 10. The storage unit 13 is a hard disk drive (HDD) that can store data in a non-volatile manner. Note that the storage unit 13 may be a semiconductor memory such as a flash memory. The communication unit 14 is an interface for communicating with the image processing device 20. The communication unit 14 may be connected to the image processing device 20 by wire or wirelessly. The communication unit 14 may also be connected to the image processing device 20 via a network such as the Internet.

The camera 11 includes an optical device that guides visible light input from the outside to each pixel position, and a detection unit that detects the amount of incident light of each RGB color at each pixel position. The light collected by the optical device is imaged on the detection unit. The detection unit has photoelectric conversion elements arranged two-dimensionally, and photoelectrically converts the received light to output a captured image consisting of the pixel values of multiple pixels arranged two-dimensionally. The captured image may be a color image consisting of the brightness of each RGB color, or a monochrome image consisting of the brightness of one color. The photoelectric conversion elements are CMOS (Complementary Metal Oxide Semiconductor) sensors, CCD (Charge Coupled Device) sensors, etc.

The captured image output from the camera 11 is stored in the memory unit 13. The control unit 12 outputs the captured image temporarily stored in the memory unit 13 to the image processing device 20 via the communication unit 14 at an appropriate timing. The captured image may be a still image obtained by capturing an image of the vehicle at a certain point in time, or a video obtained by capturing an image of the vehicle for a continuous period of time.

When the captured images are still images, the imaging device 10 may output still images at a predetermined time interval. When the captured images are moving images, the frame rate and the predetermined time interval may be such that at least a plurality of frames (e.g., 3 to 6 frames) can be obtained for each vehicle passing within the field of view, and may be changeable according to the installation position of the imaging device 10 and the expected traveling speed of the vehicle within the field of view of the camera 11.

Fig. 3 is a block diagram showing an example of the hardware configuration of an image processing device. Referring to Fig. 3, the image processing device 20 is a computer that performs arithmetic processing, and includes a CPU 201 for controlling the entire image processing device 20, a ROM (Read Only Memory) 202 that stores programs executed by the CPU 201, a RAM (Random Access Memory) 203 used as a working area for the CPU 201, a HDD 204 that stores data in a non-volatile manner, a communication unit 205 that connects the CPU 201 to a network, a display unit 206 that displays information, an operation unit 207 that accepts input of user operations, and an external storage device 210.

The communication unit 205 is an interface for connecting the image processing device 20 to a network. The network includes the Internet. Thus, the CPU 201 can communicate with a computer connected to the Internet via the communication unit 205. The communication unit 205 is also an interface for communicating with an external device. In addition to the imaging device 10, the external device to which the communication unit 205 is communicatively connected may be, for example, a warning device that warns the driver of an overloaded vehicle, an operation control device for a barrier that blocks the passage of overloaded vehicles, or a monitoring device for a guard of overloaded vehicles.

The CPU 201 downloads a program from a computer connected to a network such as the Internet and stores it in the HDD 204. When a computer connected to the network writes a program to the HDD 204, the program is stored in the HDD 204. The CPU 201 loads the program stored in the HDD 204 into the RAM 203 and executes it.

The external storage device 210 is equipped with a CD-ROM (Compact Disk Read Only Memory) 211. In this embodiment, an example will be described in which the CPU 201 executes a program stored in the ROM 202 or the HDD 204. The CPU 201 may control the external storage device 210 to read a program to be executed by the CPU 201 from the CD-ROM 211, and store the read program in the RAM 203 for execution.

The recording medium for storing the programs executed by CPU 201 is not limited to CD-ROM 211, but may be a flexible disk, cassette tape, optical disk (MO (Magnetic Optical Disc)/MD (Mini Disc)/DVD (Digital Versatile Disc)), IC card, optical card, mask ROM, EPROM (Erasable Programmable ROM) or other semiconductor memory medium. The programs referred to here include not only programs that can be executed directly by CPU 201, but also source programs, compressed programs, encrypted programs, etc.

HDD204 stores vehicle information in addition to the programs executed by CPU201. The programs stored in HDD204 include an overload detection program. Non-volatile memory such as a flash memory may be used instead of HDD204. The programs and vehicle information may also be stored in ROM202.

The vehicle information includes a specifications database and a judgment criterion value. The specifications database includes data used to identify the type of tire and the type of vehicle. Specifically, the specifications database includes the diameter of the tire mounted on the vehicle and the diameter of the rim mounted on the vehicle. The specifications database may also include the number of axles of the vehicle, the number of tires, the type of tire, the dry weight of the vehicle, and the like. The judgment criterion value is, for example, a deformation judgment amount for a maximum vehicle weight predetermined for the vehicle. The deformation judgment amount is a value indicating the degree to which the tire deforms due to differences in vehicle weight. The judgment criterion value is a threshold value used to determine whether or not the vehicle is overloaded, and is a value determined for each vehicle and tire type. Details of the deformation judgment amount and the judgment criterion value will be described later.

HDD 204 may further store captured images and their analysis results. The captured images and their analysis results are overwritten and updated in order, starting with the oldest data, within a specified capacity. In addition, the captured images and their analysis results may be erased a specified time after they are stored in HDD 204, and new captured images and their analysis results may be stored in the freed area. In particular, captured images that are determined to be overloaded may be held until they are separately obtained from the outside via communication unit 23 or the like.

Figure 4 is a diagram showing an example of functions of the CPU included in the image processing device in this embodiment. The functions shown in Figure 4 are functions realized by the CPU 201 included in the image processing device 20 by executing an overload detection program stored in the ROM 202, HDD 204, or CD-ROM 211. With reference to Figure 4, the CPU 201 includes an acquisition unit 251 that acquires an image of a subject captured by the camera 11 from the imaging device 10, a vehicle information detection unit 253 that detects information about the subject vehicle, an edge image generation unit 265 that generates an edge image from the captured image, a condition determination unit 267 that determines conditions for extracting a circle or ellipse from the edge image, a shape extraction unit 261 that extracts the shape of a circle or ellipse from the edge image, an identification unit 255 that identifies a tire portion corresponding to the tire of the subject vehicle in the captured image, a determination unit 257 that determines whether the vehicle state is overloaded, and an output unit 259 that notifies when it is determined that the vehicle is overloaded.

The acquisition unit 251 acquires from the imaging device 10 the captured image output by the camera 11 of the imaging device 10, which captures an image of a subject. The imaging device 10 captures an image of a vehicle as a subject while the vehicle passes through the imaging range. Therefore, since the imaging device 10 captures an image of the vehicle as a subject, the captured image includes a portion corresponding to the vehicle as the subject. The acquisition unit 251 outputs the captured image acquired from the imaging device 10 to the vehicle information detection unit 253, the identification unit 255, the edge image generation unit 265, and the condition determination unit 267. The captured image may be a video or a still image. When the captured image is a video, the acquisition unit 251 outputs one or more still images selected from a plurality of frames included in the video.

The vehicle information detection unit 253 detects vehicle information related to the vehicle. For example, the vehicle information detection unit 253 extracts a portion corresponding to the license plate of the vehicle included in the captured image, and extracts information written on the license plate by character recognition of the extracted portion. The vehicle information detection unit 253 then searches for vehicle information stored in advance in the HDD 204 using the information written on the license plate, and acquires vehicle information of the vehicle included as a subject in the captured image. The vehicle model may also be detected from the portion of the vehicle included in the captured image. The vehicle information includes a specification database and a judgment reference value. The specification database includes the diameter of the tires mounted on the vehicle and the diameter of the rims mounted on the vehicle. The judgment reference value is a deformation judgment amount for a maximum vehicle weight predetermined for the vehicle. The vehicle information detection unit 253 outputs the detected vehicle information to the condition determination unit 267 and the judgment unit 257.

In this embodiment, an example in which vehicle information is stored in HDD 204 will be described, but the vehicle information may be stored in a computer other than the image processing device 20. For example, when vehicle information is stored in a server connected to the image processing device 20 via a network so as to be able to communicate with the image processing device 20, the vehicle information detection unit 253 transmits information written on the license plate to the server and acquires vehicle information of the subject vehicle from the server. The vehicle information specification database and the judgment criterion value may be stored in a separate computer. For example, the specification database may be stored in a server, and the judgment criterion value may be stored in the image processing device 20.

The edge image generating unit 265 generates an edge image by performing image processing on the captured image input from the acquiring unit 251. The edge image is an image that includes edges contained in the captured image. The edge image generating unit 265 includes a first processing unit 281 that generates a first edge image by performing a first image processing on the captured image, and a second processing unit 283 that generates a second edge image by performing a second image processing on the captured image. The first processing unit 281 outputs the first edge image to the shape extracting unit 261, and the second processing unit 283 outputs the second edge image to the shape extracting unit 261.

The first image processing performed by the first processing unit 281 and the second image processing performed by the second processing unit 283 each include a smoothing process that smooths the captured image, a differential image generation process that generates a differential image by differentiating the smoothed image, and a binarization process that binarizes the differential image. The smoothing process is a process that reduces the difference in pixel value between adjacent pixels for each of multiple pixels in an image. For example, the smoothing process is a process that performs a convolution operation on the image to be processed using a smoothing filter. When different smoothing filters are used, the degree of smoothing differs.

The differential image generation process is a process that generates a differential image that indicates the difference in pixel value between adjacent pixels for each of multiple pixels in an image. For example, the differential image generation process is a process that performs a convolution operation on the image to be processed using a differential filter. The differential filter may be a first-order differential filter or a second-order differential filter.

The binarization process is a process of binarizing the differential image generated by the differential image generation process. For example, the binarization process is a process of comparing the pixel values of multiple pixels in the differential image with a binarization threshold value to distinguish between edge pixels and non-edge pixels. If the binarization threshold value differs, the degree of binarization differs.

Note that, here, an example is described in which the first image processing and the second image processing include a smoothing process, but instead of or in addition to the smoothing process, a reduction process for reducing the size of the image may be included. The reduction process can achieve the same effect as the smoothing process.

The first image processing and the second image processing are different processes. FIG. 5 is a diagram showing the difference in parameters used in the first image processing and the second image processing. Referring to FIG. 5, the smoothing process in the first image processing and the smoothing process in the second image processing differ in the degree of smoothing. The smoothing process in the first image processing has a stronger degree of smoothing than the smoothing process in the second image processing. Since smoothing is a process that reduces the difference in pixel values of adjacent pixels, the stronger the degree of smoothing, the less noise is contained in the captured image. Conversely, the lower the degree of smoothing, the less noise is contained in the captured image.

The binarization process in the first image processing and the binarization process in the second image processing have different binarization thresholds. The threshold of the binarization process in the first image processing is lower than the threshold of the binarization process in the second image processing. A differential image shows the difference in pixel value between adjacent pixels. Therefore, the lower the threshold of the binarization process, the smaller the difference in pixel value between adjacent pixels is extracted. Therefore, the first image processing detects edges with smaller changes in pixel value than the second image processing. Conversely, the higher the threshold of the binarization process, the larger the difference in pixel value between adjacent pixels is extracted. Therefore, the second image processing detects edges with larger changes in pixel value than the first image processing.

Since the degree of smoothing in the first image processing is stronger than the degree of smoothing in the second image processing, the first differential image generated by differentiating the image after the smoothing processing in the first image processing has less noise than the second differential image generated by differentiating the image after the smoothing processing in the second image processing. Therefore, it is possible to set the binarization threshold in the first image processing to a value lower than the binarization threshold in the second image processing. As a result, edges with small changes in brightness in the captured image are extracted from the first differential image generated by the first image processing. Therefore, the first edge image obtained by binarizing the first differential image by the first processing unit 281 includes edges with smaller changes in brightness than the second edge image obtained by binarizing the second differential image by the second processing unit 283.

In addition, the second differential image generated by the second image processing has more noise than the first differential image generated by the first image processing, but since the binarization threshold in the second image processing is higher than the binarization threshold in the first image processing, it is possible to prevent noise from being erroneously detected as an edge, and only edges with large changes in brightness in the captured image are extracted from the first differential image generated by the second image processing. Therefore, the second edge image generated by the second processing unit 283 includes edges with larger changes in brightness than the first edge image generated by the first processing unit 281.

In the captured image, the vehicle tire often has a small difference in brightness from the surrounding area of the tire. On the other hand, the vehicle rim is surrounded by tires, so there is often a large difference in brightness between the rim and the surrounding tires. For this reason, the first image processing generates a first edge image suitable for extracting a portion corresponding to the outer edge of the vehicle tire from the captured image, and the second image processing generates a second edge image suitable for extracting a portion corresponding to the outer edge of the vehicle rim from the captured image.

Returning to FIG. 4, the shape extraction unit 261 extracts a circular or elliptical shape from the edge image according to the conditions determined by the condition determination unit 267. The shape extraction unit 261 includes a first extraction unit 271 that extracts a first circular or elliptical shape from the first edge image input from the first processing unit 281, and a second extraction unit 273 that extracts a second circular or elliptical shape from the second edge image input from the second processing unit 283.

The condition determination unit 267 includes a first setting unit 285 and a second setting unit 287. The first setting unit 285 sets a first extraction condition for the first extraction unit 271 to extract a portion corresponding to the outer edge of the tire of the vehicle in the captured image as a first circle or ellipse based on the vehicle information. Specifically, the first setting unit 285 determines the first diameter information and a threshold value of the voting value in the Hough transform. The first diameter information is a diameter within a predetermined range from the diameter of the tire part of the vehicle in the captured image. The first setting unit 285 determines the diameter of the tire part of the vehicle in the captured image from the ratio between the length of the reference part of the vehicle determined by the vehicle information and the diameter of the tire of the vehicle. The reference part of the vehicle may be previously determined as a part that is easy to extract from the captured image. For example, the reference part of the vehicle may be the height of the vehicle or the length of the vehicle. When the reference part of the vehicle is the vehicle height, the first setting unit 285 determines the diameter of the tire part included in the captured image from the height of the vehicle part included in the captured image and the ratio of the vehicle height and tire diameter determined in the vehicle information. The first setting unit 285 sets the first diameter information as a condition for extracting a first circle or ellipse that corresponds to the outer edge of the tire part.

The voting value cast for each pixel in the Hough transform increases as the radius of the circle or ellipse increases. For this reason, the condition determination unit 267 determines a threshold value according to the first diameter information. The threshold value is determined using a table or an arithmetic expression that predetermines the relationship between the first diameter information and the threshold value. The first setting unit 285 sets the first diameter information to the diameter of the first circle or ellipse to be extracted by the Hough transform, and sets the threshold value determined for the diameter as the threshold value used for judgment in the Hough transform.

Here, the center of the ellipse is the point where the major axis and minor axis of the ellipse intersect. Depending on the relative positions of the tire and camera 11, the tire portion in the captured image may be a circle or an ellipse. When the optical axis of camera 11 coincides with the direction of the axle passing through the center of the tire, the tire portion in the captured image will be a circle, and when the optical axis of camera 11 intersects with the direction of the axle passing through the center of the tire, the tire portion will be an ellipse. When camera 11 is installed so that the optical axis of camera 11 is at the same height as the axle passing through the center of the tire, the optical axis of camera 11 and the axle passing through the center of the tire will intersect in the horizontal direction. In this case, the tire portion in the captured image will be an ellipse, but the vertical length will not change. For this reason, the diameter of the ellipse can be taken as the major axis.

The second setting unit 287 sets a second extraction condition for the second extraction unit 273 to extract a portion corresponding to the outer edge of the vehicle rim in the captured image as a second circle or ellipse based on the vehicle information. Specifically, the second setting unit 287 determines the second diameter information and a threshold value of the voting value in the Hough transform. The second diameter information is a diameter within a predetermined range from the diameter of the rim portion of the vehicle in the captured image. The second setting unit 287 determines the diameter of the rim portion of the vehicle in the captured image from the ratio between the length of the reference portion of the vehicle defined by the vehicle information and the diameter of the rim of the vehicle. When the reference portion of the vehicle is the height of the vehicle, the second setting unit 287 determines the diameter of the rim portion included in the captured image from the height of the portion of the vehicle included in the captured image and the ratio between the height of the vehicle defined by the vehicle information and the diameter of the rim. The second setting unit 287 sets the second diameter information to the diameter of the second circle or ellipse to be extracted by the Hough transform, and sets the threshold value determined for the second diameter information to the threshold value used for judgment in the Hough transform. The threshold value is determined using a table or an arithmetic expression that defines a relationship between the second diameter information and the threshold value in advance. The second setting unit 287 sets the second diameter information to the diameter of the second circle or ellipse to be extracted by the Hough transform, and sets the threshold value determined for the diameter to the threshold value used for judgment in the Hough transform.

The first setting unit 285 may correct the threshold value according to the diameter of the tire portion included in the captured image, and the second setting unit 287 may correct the threshold value according to the diameter of the rim portion included in the captured image. The tire portion in the captured image does not necessarily include the entire outer edge of the tire, and similarly, the rim portion in the captured image does not necessarily include the entire outer edge of the rim. For this reason, the threshold value may be corrected using a correction value determined by experiment or the like. The correction value is preferably determined based on the degree of smoothing processing and the threshold value of the binarization processing.

Figure 6 is a diagram showing an example of a first extraction condition and a second extraction condition. Referring to Figure 6, the first extraction condition has a larger radius of the circle or ellipse than the second extraction condition, and a smaller threshold than the second extraction condition.

Returning to FIG. 4, the shape extraction unit 261 includes a first extraction unit 271 and a second extraction unit 273. The first extraction unit 271 extracts a first circle or ellipse by performing a Hough transform on the first edge image input from the first processing unit 281 according to the first extraction condition set by the first setting unit 285. The first extraction condition is a value for extracting a tire portion corresponding to the tire of the vehicle, which is the subject, from the captured image, so the first circle or ellipse is extracted as a candidate for the outer edge of the tire portion included in the captured image. There are cases where multiple first circles or ellipses are extracted. The first extraction unit 271 outputs a pair of the center coordinates and radius of the first circle or ellipse to the identification unit 255.

The second extraction unit 273 extracts a second circle or ellipse by performing a Hough transform on the second edge image input from the second processing unit 283 according to the second extraction conditions set by the second setting unit 287. Since the second extraction conditions are values for extracting a rim portion corresponding to the rim of the vehicle, which is the subject, from the captured image, the second circle or ellipse is extracted as a candidate for the outer edge of the rim portion included in the captured image. Multiple second circles or ellipses may be extracted. The second extraction unit 273 outputs a pair of the center coordinates and radius of the second circle or ellipse to the identification unit 255.

The identification unit 255 receives one or more pairs of the center coordinates and radius of the first circle or ellipse from the first extraction unit 271, and one or more pairs of the center coordinates and radius of the second circle or ellipse from the second extraction unit 273. The identification unit 255 identifies the tire part in the captured image. Specifically, the identification unit 255 determines a pair of a first circle or ellipse and a second circle or ellipse whose center coordinates are within a predetermined range as a target pair. The rotation axis of the tire and the rotation axis of the rim coincide. Therefore, the center of the first circle or ellipse including the outer edge of the tire part and the center of the second circle or ellipse including the outer edge of the rim part should coincide in the captured image. Therefore, there is a high probability that a pair of two circles or ellipses whose center coordinates are within a predetermined range corresponds to a pair of a circle or ellipse including the outer edge of the tire part and a circle or ellipse including the outer edge of the rim part.

Furthermore, the identification unit 255 determines a pair of two circles or ellipses whose radius ratio is within a predetermined range. The radius ratio between the tire and the rim is determined by the vehicle. Therefore, when there are multiple circles or ellipses whose center coordinate distances are within a predetermined range, it is possible to identify a pair of two circles or ellipses whose radius ratio is within a predetermined range. Therefore, when a vehicle is identified, it is possible to more accurately extract a pair of two circles or ellipses that respectively correspond to the outer edge of the tire portion and the outer edge of the rim portion. Furthermore, the identification unit 255 determines the radius ratio based on the respective radii of the tire and rim defined in the vehicle information. Therefore, the radius ratio is accurately determined.

The identification unit 255 identifies an edge included within a predetermined range from the first circle or ellipse included in the target set in the captured image as part of the outer edge of the tire part. The identification unit 255 also determines an edge continuous with an edge constituting a part of the outer edge of the identified tire part as part of the outer edge of the tire part. The part of the tire that comes into contact with the ground may not be included within a predetermined range from the first circle or ellipse in the captured image. Since the part of the tire that does not come into contact with the ground is included within a predetermined range from the first circle or ellipse in the captured image, the outer edge of the tire part for the part of the tire that comes into contact with the ground is determined by setting an edge continuous with an edge corresponding to the part of the tire that does not come into contact with the ground as the outer edge of the tire part. The identification unit 255 outputs the outer edge of the tire part in the captured image to the determination unit 257. Note that an edge parallel to the road surface in the captured image within the first circle or ellipse may be determined as the outer edge of the tire part.

Figure 7 is a diagram showing an example of the relationship between the tire part included in the captured image and the circle or ellipse extracted by the Hough transform. Referring to Figure 7, the outer edge of the tire part included in the captured image and the outer edge of the rim part are shown by thin solid lines. The circle extracted by the Hough transform is shown by dotted lines or dashed lines. In addition to a first circle shown by a thin dotted line for the outer edge of the tire part and a second circle shown by a thick dotted line for the outer edge of the rim part, two circles shown by thick dashed lines are included as circles extracted by the Hough transform. Of these four circles, the pair of two circles whose center coordinates are within a predetermined range are the first circle corresponding to the outer edge of the tire part shown by the thin dotted line and the second circle corresponding to the outer edge of the rim part shown by the thick dotted line. Note that the two circles shown by the dashed lines are circles that are erroneously detected by the Hough transform.

Returning to FIG. 4, the determination unit 257 receives the outer edge of the tire portion in the captured image from the identification unit 255 and the vehicle information from the vehicle information detection unit 253. The determination unit 257 determines whether the vehicle is overloaded or not based on the shape of the outer edge of the tire portion and the vehicle information. For example, the determination unit 257 determines the center, radius, and deformation judgment amount of the tire portion based on the outer edge of the tire portion, and determines whether the deformation judgment amount exceeds a judgment reference value corresponding to the vehicle's load limit weight. If the deformation judgment amount exceeds the judgment reference value, the determination unit 257 determines that the vehicle is overloaded, and outputs a warning instruction to the output unit 259.

In response to the input of a warning instruction, the output unit 259 outputs warning information. For example, the output unit 259 displays a warning message on the display unit 206. A warning sound may also be generated. Furthermore, a gate may be closed to prevent the passage of vehicles.

8 is a diagram for explaining the deformation determination amount of a tire. Referring to FIG. 8, a tire T is attached to a rim R of a wheel of a wheel. Although the tire T has a structure such as a recess on its surface, to be precise, the following description will be given assuming that a circle including the outer edge of the tire is detected from a front image of the tire T viewed from the rotation axis direction, ignoring the structure such as the recess. Here, the horizontal direction is the x-axis direction, and the vertical direction is the y-axis direction. The road surface on which the tire touches the ground is assumed to be substantially horizontal and extends along the x-axis direction. Although not particularly limited, the following description will be given assuming that the shooting range of the camera 11 is a rectangular area, and the vertical and horizontal directions of the captured image are respectively along the y-axis direction and the x-axis direction. The origin position of each axis may be arbitrarily determined, for example, at the lower left end of the image or the center position of the image.

The tire T has a diameter L2t when not deformed. The tire T is deformed into a planar shape along the road surface by the weight of the vehicle, and is compressed vertically by a compression width dLy. In the front image, the part where the tire T comes into contact with the road surface is called the ground line. If the tire air pressure, material, and structure are constant, the ground width Lcx between both ends of the ground line becomes longer as the weight of the vehicle including the load weight increases. Accordingly, the distance Lcy from the rotation axis position OR, which is the center of rotation of the tire T, to the road surface and the vertical width L2cy of the tire also become shorter. In other words, the shape of the tire T is a combination of the ground line whose bottom surface is linear over the ground width Lcx and the arc portion that does not come into contact with the ground and does not deform. Therefore, if the parameters indicating the ground portion and the parameters indicating the arc portion are obtained, the shape of the tire T can be specified. On the other hand, the rim R does not deform from a perfect circle shape due to the weight of the vehicle, so the rim radius Lr does not change in the x-axis direction and the y-axis direction.

To determine whether or not an overload exists, the contact width Lcx is set as the deformation value as a parameter indicating the contact portion of the tire T, the radius Lth of the tire T is set as the tire reference value as a parameter indicating the arc portion of the tire T, and the ratio of the deformation value to the tire reference value is set as the deformation judgment amount. The greater the weight of the vehicle, the greater the deformation judgment amount. The presence or absence of an overload is determined by determining whether or not the deformation judgment amount of the tire T exceeds the judgment reference value corresponding to the load limit weight set for the vehicle. The load limit weight set for a vehicle is the maximum weight set in advance for that vehicle.

Note that other combinations of deformation values and tire reference values may be used as long as they are a combination of a value that changes with vehicle weight and a value that does not change with vehicle weight.

Figure 9 is a flow chart showing an example of the flow of overload detection processing. The overload detection processing is a process executed by the CPU 201 of the image processing device 20 by executing an image processing program stored in the ROM 202, the HDD 204, or the CD-ROM 211. Referring to Figure 9, the CPU 201 of the image processing device 20 controls the imaging device 10 to cause the camera 11 to capture an image of the vehicle as a subject (step S01). In the next step S02, an image obtained by the camera 11 capturing an image of the vehicle is acquired from the imaging device 10, and the process proceeds to step S03. Then, vehicle information is detected. Based on the information written on the license plate of the vehicle included in the captured image, vehicle information corresponding to the vehicle is acquired from the vehicle information stored in the HDD 204.

In step S04, a tire portion extraction process is executed, and the process proceeds to step S05. The tire portion extraction process will be described in detail later, but this process involves extracting the portion of the captured image that corresponds to the vehicle's tires as the tire portion from the captured image.

In step S05, it is determined whether the vehicle is overloaded. Whether the vehicle is overloaded is determined based on the shape of the outer edge of the tire portion extracted in step S04 and the vehicle information. Specifically, based on the outer edge of the tire portion, the deformation judgment amount is determined from the radius Lth of the tire portion and the contact width Lcx, and it is determined whether the deformation judgment amount exceeds a judgment reference value corresponding to the vehicle's load limit weight. If the deformation judgment amount exceeds the judgment reference value, it is determined that the vehicle is overloaded, and the process proceeds to step S06, but if not, the process ends.

In step S06, a warning is output. For example, a warning message is displayed on the display unit 206. A warning sound may also be generated. Furthermore, the gate may be closed to prevent the passage of vehicles.

Figure 10 is a flow chart showing an example of the flow of the tire portion extraction process. The tire portion extraction process is a process executed in step S04 of the overload detection process. Referring to Figure 10, the CPU 201 determines a first condition (step S11). The first condition is a condition that is set for detecting the outer edge of the tire portion included in the captured image. The first condition includes parameters for image processing and parameters used in the Hough transform. The image processing parameters include a parameter that determines the degree of smoothing processing and a parameter that determines the threshold value for the binarization processing. The parameters used in the Hough transform include the diameter of the circle or ellipse and a threshold value that is compared with the voting value. In addition, a correction value for correcting the threshold value may be determined.

In step S12, a first edge extraction process is performed, and the process proceeds to step S13. In the first edge extraction process, edge extraction process (described later) is performed using the image processing parameters determined as the first condition in step S11. The edge extraction process includes smoothing process, differential image generation process, and binarization process, and is a process for converting the captured image into a first edge image.

In step S13, a first Hough transform is performed, and the process proceeds to step S14. The first edge image generated in step S12 is treated as the processing target, and a Hough transform is performed using the parameters used for the Hough transform determined as the first condition in step S11.

In step S14, a second condition is determined. The second condition is a condition determined for detecting the outer edge of the rim portion included in the captured image. The second condition includes image processing parameters and parameters used in the Hough transform. The image processing parameters include a parameter that determines the degree of smoothing processing and a parameter that determines the threshold value of the binarization processing. The smoothing processing parameters included in the second condition are parameters that provide a lower degree of smoothing than the parameters in the first condition. The binarization processing parameters included in the second condition are parameters that provide a higher degree of binarization than the parameters in the first condition, in other words, parameters with larger values.

In step S15, a second edge extraction process is performed, and the process proceeds to step S16. In the second edge extraction process, edge extraction process (described later) is performed using the image processing parameters determined as the second condition in step S14. The edge extraction process includes a smoothing process, a differential image generation process, and a binarization process, and is a process for converting the captured image into a second edge image.

In step S16, a second Hough transform is performed, and the process proceeds to step S17. The second edge image generated in step S16 is treated as the processing target, and a Hough transform is performed using the parameters used for the Hough transform determined as the second condition in step S14.

In step S17, the first shape is extracted. By performing the Hough transform in step S13, candidates for a circle or ellipse that include the outer edge of the tire portion included in the captured image are identified, and the first shape is extracted based on the edges included in the area within the identified circle or ellipse in the captured image. The outer edge of the tire portion is the subject of extraction, and the portion where the tire contacts the ground is a horizontal straight line, so a shape composed of edges in the captured image that overlap with the identified circle or ellipse and horizontal edges located near and above the lowest point of the identified circle or ellipse is extracted as the first shape.

In step S18, the second shape is extracted, and the process proceeds to step S19. By performing the Hough transform, candidates for a circle or ellipse that include as part of the outer edge of the rim portion contained in the captured image are identified, and the second shape is extracted based on the edges contained in the area within the identified circle or ellipse in the captured image. Since the outer edge of the rim is the subject of extraction, a shape formed by edges that overlap with the identified circle or ellipse is extracted as the second shape.

In step S19, the tire portion in the captured image is identified, and the process returns to the overload detection process. Specifically, the portion of the captured image surrounded by the first shape extracted in step S17 and the second shape extracted in step S18 is identified as the tire portion. In step S17, multiple candidates that are circles or ellipses that include the outer edge of the tire portion included in the captured image may be identified, and in step S18, multiple candidates that are circles or ellipses that include the outer edge of the rim portion included in the captured image may be identified. In this case, among one or more candidates identified as the first shape and one or more candidates identified as the second shape, pairs of first shape candidates and second shape candidates whose center distances are within a predetermined range are determined as the first shape and the second shape, respectively. Furthermore, when there are multiple pairs of first shape candidates and second shape candidates whose center coordinate distances are within a predetermined range, pairs of first shape candidates and second shape candidates whose radius ratios are within a predetermined range are determined. Therefore, when a vehicle is identified, a set of two circles or ellipses corresponding to the outer edges of the tire portion and the rim portion, respectively, can be extracted more accurately. The radius ratio is determined based on the respective radii of the tire and the rim, which are determined in the vehicle information.

Figure 11 is a flowchart showing an example of the flow of edge extraction processing. The edge extraction processing is a process executed in steps S12 and S15 of the tire portion extraction processing. Referring to Figure 11, the CPU 201 executes a smoothing process on the captured image (step S21) and proceeds to step S22. When step S12 of the tire portion extraction processing is executed, the first condition is set, so the smoothing process is executed using a parameter indicating the degree of smoothing determined by the first condition, here the first smoothing filter. When step S15 of the tire portion extraction processing is executed, the second condition is set, so the smoothing process is executed using a parameter indicating the degree of smoothing determined by the second condition, here the second smoothing filter. The first smoothing filter has a stronger degree of smoothing than the second smoothing filter.

In step S22, a differential image generation process is executed, and the process proceeds to step S23. The smoothed captured image is subjected to a convolution operation using a differential filter to generate a differential image.

In step S23, binarization processing is performed, and processing returns to the tire portion extraction processing. When step S12 of the tire portion extraction processing is performed, the first condition is set, so the binarization processing is performed using a parameter indicating the degree of binarization determined by the first condition, here the first threshold value. When step S15 of the tire portion extraction processing is performed, the second condition is set, so the binarization processing is performed using a parameter indicating the degree of binarization determined by the second condition, here the second threshold value. The first threshold value provides a lower degree of binarization than the second threshold value. In other words, the first threshold value is smaller than the second threshold value.

As described above, in the overload detection system 1 in this embodiment, the image processing device 20 includes an acquisition unit 251 that acquires an image obtained by photographing the side of the vehicle, a first processing unit 281 that performs a first image processing on the captured image and extracts edges contained in the image on which the first image processing has been performed, a second processing unit 283 that performs a second image processing different from the first image processing on the captured image and extracts edges contained in the image on which the second image processing has been performed, a first extraction unit 271 that extracts a first circle or ellipse by performing a Hough transform on the edges extracted by the first processing unit 281, a second extraction unit 273 that extracts a second circle or ellipse by performing a Hough transform on the edges extracted by the second processing unit 283, and an identification unit 255 that identifies the tire portion in the captured image corresponding to the vehicle's tires based on the first circle or ellipse and the second circle or ellipse. Since the first image processing and the second image processing are different, the edges extracted from the image obtained by performing the first image processing on the captured image may be the same as or different from the edges extracted from the image obtained by performing the second image processing on the captured image. If the first image processing is a process in which the portion corresponding to the outer edge of the vehicle's tire is likely to appear as an edge in the captured image, a first circle or ellipse including the portion corresponding to the outer edge of the tire is likely to be extracted. Also, if the second image processing is a process in which the portion corresponding to the outer edge of the vehicle's rim is likely to appear as an edge in the captured image, a second circle or ellipse including the portion corresponding to the outer edge of the rim is likely to be extracted. Therefore, the tire portion corresponding to the vehicle's tire can be accurately extracted from the captured image obtained by capturing an image of the vehicle.

In addition, in the image processing performed by the image processing device 20, the degree of smoothing in the smoothing process included in the first image processing is greater than the degree of smoothing in the smoothing process included in the second image processing, and the binarization threshold in the binarization process included in the first image processing is smaller than the binarization threshold in the binarization process included in the second image processing. Therefore, the differential image generated by the first image processing has less noise than the differential image generated by the second image processing, and edges with small changes in brightness in the captured image can be extracted from the differential image generated by the first image processing. At this time, since the differential image generated by the first image processing has less noise than the differential image generated by the second image processing, it is possible to prevent noise from being erroneously detected as an edge. Furthermore, the differential image generated by the second image processing has more noise than the differential image generated by the first image processing, but because the binarization threshold in the second image processing is larger than the binarization threshold in the first image processing, it is possible to prevent the noise from being erroneously detected as an edge, and it is possible to extract only edges with large changes in brightness from the captured image from the differential image generated by the second image processing.

In addition, in the image processing performed by the image processing device 20, a reduction process may be performed instead of or in addition to the smoothing process.

The image processing device 20 also sets the first diameter information determined based on the size of the tire mounted on the vehicle as the first condition, and sets the second diameter information determined based on the size of the rim mounted on the vehicle as the second condition. This allows the portion corresponding to the outer edge of the tire to be extracted as a first circle or ellipse, and the portion corresponding to the outer edge of the rim to be extracted as a second circle or ellipse.

The image processing device 20 also sets a first threshold determined by the first diameter information as a first condition to be compared with the voting result in the Hough transform, and sets a second threshold determined by the second diameter information as a second condition to be compared with the voting result in the Hough transform. Therefore, a first circle or ellipse corresponding to the diameter of the outer edge of the tire and a second circle or ellipse corresponding to the diameter of the outer edge of the rim can be extracted from the captured image.

In addition, the image processing device 20 determines whether the vehicle is overloaded based on the tire portion identified from the captured image, making it easy to determine whether the vehicle is overloaded.

In this embodiment, vehicle information is detected from the captured image, but the first circle or ellipse and the second circle or ellipse may be extracted without detecting vehicle information. In this case, the first diameter information and the second diameter information including the tire diameter and the rim diameter are not provided as parameters for performing the Hough transform, but one or more first circles or ellipses and one or more second circles or ellipses are extracted. In this case, the first circle or ellipse and the second circle or ellipse may be identified based on the distance between their centers. For example, the pair with the shortest distance between their centers may be identified.

In this embodiment, an example is shown in which the camera 11 outputs a captured image obtained by photoelectrically converting visible light, but it may also be configured to output a captured image obtained by photoelectrically converting infrared light. As the temperature of the tires increases when the vehicle is traveling, it becomes easier to identify the tire part by extracting from the captured image a part where infrared light is detected at a predetermined value or more.

The embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects. The scope of the present invention is indicated by the claims, not by the above description, and is intended to include all modifications within the meaning and scope of the claims.

1 Overload detection system, 10 Imaging device, 11 Camera, 12 Control unit, 13 Storage unit, 14 Communication unit, 20 Image processing device, 23 Communication unit, 201 CPU, 202 ROM, 203 RAM, 204 HDD, 205 Communication unit, 206 Display unit, 207 Operation unit, 210 External storage device, 211 CD-ROM, 251 Acquisition unit, 253 Vehicle information detection unit, 255 Identification unit, 257 Judgment unit, 259 Output unit, 261 Shape extraction unit, 265 Edge image generation unit, 267 Condition determination unit, 271 First extraction unit, 273 Second extraction unit, 281 First processing unit, 283 Second processing unit, 285 First setting unit, 287 Second setting unit.

Claims (9)

An acquisition means for acquiring an image obtained by photographing a side of a vehicle;
a first processing means for performing a first image processing on the captured image and extracting edges included in the image that has been subjected to the first image processing;
a second processing means for performing a second image processing different from the first image processing on the captured image and extracting edges included in the image on which the second image processing has been performed;
a first extraction means for extracting a first circle or an ellipse by performing a Hough transform on the edge extracted by the first processing means;
a second extraction means for extracting a second circle or ellipse by performing a Hough transform on the edge extracted by the second processing means;
and an identification means for identifying a tire portion in the captured image that corresponds to a tire of the vehicle based on the first circle or ellipse and the second circle or ellipse. each of the first image processing and the second image processing includes a smoothing processing for smoothing an image, a differential image generating processing for generating a differential image by differentiating the smoothed image, and a binarization processing for binarizing the differential image;
a degree of smoothing in the smoothing process included in the first image processing is greater than a degree of smoothing in the smoothing process included in the second image processing,
The image processing device according to claim 1 , wherein a binarization threshold in the binarization process included in the first image processing is smaller than a binarization threshold in the binarization process included in the second image processing. each of the first image processing and the second image processing includes a reduction processing for reducing an image, a differential image generating processing for generating a differential image by differentiating the reduced image, and a binarization processing for binarizing the differential image;
a degree of reduction in the reduction process included in the first image processing is greater than a degree of reduction in the reduction process included in the second image processing,
The image processing device according to claim 1 , wherein a binarization threshold in the binarization process included in the first image processing is smaller than a binarization threshold in the binarization process included in the second image processing. The first extraction means is configured to set first diameter information determined based on a size of a tire mounted on the vehicle,
4. The image processing device according to claim 1, wherein the second extraction means sets second diameter information determined based on a size of a rim mounted on the vehicle. the first extraction means sets a first threshold determined by the first radius information as a threshold to be compared with a voting result in a Hough transform;
5. The image processing apparatus according to claim 4, wherein said first extraction means sets a second threshold determined by said second diameter information as a threshold to be compared with a voting result in a Hough transform. The image processing device according to any one of claims 1 to 5, further comprising a determination means for determining whether the vehicle is overloaded or not based on the tire portion identified by the identification means. An image processing device according to any one of claims 1 to 6,
an imaging means for imaging the vehicle and outputting the resulting image to the image processing device. An acquisition step of acquiring an image obtained by photographing a side of a vehicle;
a first processing step of performing a first image processing on the captured image and extracting edges included in the image subjected to the first image processing;
a second processing step of performing a second image processing different from the first image processing on the captured image and extracting edges included in the image subjected to the second image processing;
a first extraction step of extracting a first circle or ellipse by performing a Hough transform on the edges extracted in the first processing step;
a second extraction step of extracting a second circle or ellipse by performing a Hough transform on the edges extracted in the second processing step;
and a step of identifying a tire portion in the captured image that corresponds to a tire of the vehicle based on the first circle or ellipse and the second circle or ellipse. An acquisition step of acquiring an image obtained by photographing a side of a vehicle;
a first processing step of performing a first image processing on the captured image and extracting edges included in the image subjected to the first image processing;
a second processing step of performing a second image processing different from the first image processing on the captured image and extracting edges included in the image subjected to the second image processing;
a first extraction step of extracting a first circle or ellipse by performing a Hough transform on the edges extracted in the first processing step;
a second extraction step of extracting a second circle or ellipse by performing a Hough transform on the edges extracted in the second processing step;
and a specifying step of specifying a tire portion in the captured image corresponding to a tire of the vehicle based on the first circle or ellipse and the second circle or ellipse.

Images