Vehicle dimension estimation system and vehicle dimension estimation method

Abstract

The present invention provides a configuration that can estimate, with a simple configuration, the width and height of vehicles traveling on a road.　This dimension estimation system can be connected to a camera capable of photographing a vehicle traveling on a road, and comprises: an extraction unit that extracts a to-be-measured region of the vehicle from an image photographed by the camera; and an estimation unit that estimates the width and / or height of the vehicle on the basis of the ratio of the size of the to-be-measured region of the vehicle that was extracted and a subject in the image that was set in advance.

Description

Vehicle dimension estimation system and vehicle dimension estimation method

[0001] The present invention relates to a vehicle dimension estimation system and a vehicle dimension estimation method.

[0002] From the perspective of ensuring road safety and smooth traffic, there are systems in place that impose restrictions on the passage of vehicles that exceed a certain size and require prior permission from road administrators. For example, Japan's Vehicle Restriction Ordinance stipulates maximum limits on the width, weight, height, length, and minimum turning radius of vehicles that can travel on roads.

[0003] Furthermore, in the field of autonomous driving, etc., there is known a device that uses vehicle width and vehicle height when determining the distance between a vehicle and a vehicle captured by a camera. For example, Patent Document 1 discloses a vehicle width detection device that infers a first projected vehicle width using a learning model that learns the relationship between an image of a vehicle to be detected captured by a camera and the vehicle width (first projected vehicle width) projected onto a plane perpendicular to the direction from the principal point of the camera lens to the vehicle.

[0004] Japanese Patent Application Laid-Open No. 2019-20171

[0005] To estimate the width and height of vehicles traveling on a road using the configuration of Patent Document 1, it is necessary to prepare a huge amount of training data and generate a learning model that can infer vehicle width from all angles for a large number of vehicle types.

[0006] An object of the present disclosure is to provide a vehicle dimension estimation system and a vehicle dimension estimation method that can estimate the width and height of a vehicle traveling on a road with a simpler configuration.

[0007] According to a first aspect, there is provided a vehicle dimension estimation system that can be connected to a camera that can photograph vehicles traveling on a road, and that includes an extraction unit that extracts a measured area of ​​the vehicle from an image photographed by the camera, and an estimation unit that estimates at least one of the vehicle width or vehicle height of the vehicle based on the size ratio between the extracted measured area of ​​the vehicle and a previously set subject that appears in the image.

[0008] According to a second aspect, there is provided a method for estimating vehicle dimensions, which extracts a measured area of ​​a vehicle from an image captured by a camera capable of capturing images of a vehicle traveling on a road, and estimates at least one of the vehicle width or height of the vehicle based on the size ratio between the extracted measured area of ​​the vehicle and a previously set subject appearing in the image.

[0009] The present disclosure can contribute to providing a configuration that can estimate the width and height of a vehicle traveling on a road with a simple configuration.

[0010] FIG. 1 is a diagram showing one configuration of the present disclosure. FIG. 2 is a flowchart showing the operation of the present disclosure. FIG. 3 is a diagram for explaining a vehicle width estimation method according to the present disclosure. FIG. 4 is a diagram for explaining another vehicle width estimation method according to the present disclosure. FIG. 5 is a diagram showing one configuration of the present disclosure. FIG. 6 is a diagram showing an example of the installation position of a camera connected to the dimension estimation system of the present disclosure. FIG. 7 is another diagram showing an example of the installation position of a camera connected to the dimension estimation system of the present disclosure. FIG. 8 is another diagram showing an example of the installation position of a camera connected to the dimension estimation system of the present disclosure. FIG. 9 is a diagram for explaining guide lines for adjusting the angle of view superimposed on the camera of the present disclosure. FIG. 10 is a diagram for explaining the size of a license plate. FIG. 11 is a flowchart showing the operation of the present disclosure. FIG. 12 is a diagram showing another configuration of the present disclosure. FIG. 13 is a flowchart showing the operation of the present disclosure. FIG. 14 is a diagram showing another configuration of the present disclosure. FIG. 15 is a flowchart showing the operation of the present disclosure. FIG. 16 is a diagram showing another configuration of the present disclosure. FIG. 17 is a diagram showing the configuration of a computer constituting an information processing device of the present disclosure.

[0011] First, an overview of one embodiment of the present disclosure will be described with reference to the drawings. In this disclosure, the drawings relate to one or more embodiments. The reference numerals in the drawings attached to this overview are attached to each element for convenience as an example to facilitate understanding, and are not intended to limit the present disclosure to the illustrated form. Furthermore, connecting lines between blocks in the drawings and the like referred to in the following description include both bidirectional and unidirectional lines. Unidirectional arrows are used to schematically indicate the flow of main signals (data) and do not exclude bidirectionality. A program is executed via a computer device, which includes, for example, a processor, a storage device, an input device, a communication interface, and, if necessary, a display device. Furthermore, this computer device is configured to be able to communicate with internal or external devices (including computers) via the communication interface, whether wired or wireless. Although ports or interfaces are present at the input / output connection points of each block in the drawings, they are not shown.

[0012] In one embodiment, the present disclosure can be realized by a dimension estimation system 10 that can be connected to a camera C that can capture images of vehicles V1 to V3 traveling on a road, as shown in FIG. 1. More specifically, the dimension estimation system 10 includes an extraction unit 11 and an estimation unit 12. The extraction unit 11 extracts measurement areas of the vehicles V1 to V3 from an image captured by the camera C. The estimation unit 12 estimates at least one of the vehicle width and vehicle height of the vehicles based on a size ratio between the extracted measurement areas of the vehicles V1 to V3 and a previously set subject (subject of known size) that appears in the image.

[0013] The dimension estimation system 10 configured as described above operates as follows: First, the dimension estimation system 10 extracts the measurement areas of the vehicles V1 to V3 from the image captured by the camera C (step S01 in FIG. 2).

[0014] Next, the dimension estimation system 10 estimates at least one of the vehicle width and vehicle height of the vehicles V1 to V3 based on the size ratio between the extracted measurement area of ​​the vehicles V1 to V3 and a previously set object captured in the image (step S02 in FIG. 2). Here, the "measurement area" of the vehicle refers to an area selected for estimating the vehicle's width or height. For example, when estimating the vehicle width, a front fender or humpback is selected. When estimating the vehicle height, a front projection image of the vehicle is selected. Furthermore, a license plate or a road lane is selected as the object captured in the image. It is desirable that the "measurement area" and the "object captured in the image" are located at approximately the same distance from the camera C. This simplifies the calculation of the size ratio between the "measurement area" and the "object."

[0015] 3 and 4 are diagrams illustrating a vehicle width estimation method according to the present disclosure. FIG. 3 shows a vehicle photographed by camera C. The large dashed rectangle in FIG. 3 indicates the "measurement area." In the example of FIG. 3, a license plate is selected as the "subject of known size." The section indicated by w1 in FIG. 3 is the width of the license plate, which is known. The section indicated by w2 in FIG. 3 is the vehicle width. Dividing the number of pixels in the vehicle width w2 by the number of pixels in w1 gives the ratio of the size of the "measurement area" to the "subject." Multiplying the known license plate width by this ratio allows for accurate calculation of the vehicle width. For example, if the width of the measurement area on the image is six times the width of the license plate and the license plate width is 440 mm, the vehicle width is calculated as 6 x 440 mm = 2640 mm.

[0016] FIG. 4 shows a method for calculating vehicle width when a road lane is used as an object of known size. The section indicated by w3 in FIG. 4 is the lane width, which is known. The section indicated by w2 in FIG. 4 is the vehicle width. Dividing the number of pixels of this vehicle width w2 by the number of pixels of w3 gives the ratio of the size of the "measurement area" to the "object." Multiplying the known lane width by this ratio allows for accurate calculation of vehicle width. For example, if the width of the measurement area on the image is 0.8 times the lane width and the lane width is 3 m, the vehicle width is calculated as 0.8 x 3 m = 2.4 m.

[0017] As a method for calculating the vehicle width using an image, it is also possible to calculate the vehicle width directly from the number of pixels by determining the length per pixel that constitutes the image in advance. However, with this method, the distance between the camera C and the vehicle V may change depending on the timing of the image capture, which may result in insufficient accuracy. According to the method disclosed herein, the vehicle width is calculated by comparison with a subject of known size, which has the advantage of providing sufficient accuracy even if the distance between the camera C and the vehicle V changes.

[0018] 3 and 4, the vehicle width is estimated, but the vehicle height can be estimated using a similar method. Furthermore, the dimension estimation system 10 may estimate both the vehicle width and the vehicle height. By estimating both the vehicle width and the vehicle height, the dimension estimation system 10 can also estimate the so-called vehicle clearance (projected size from the front, front dimension).

[0019] As described above, according to the present disclosure, it is possible to estimate the width and height of a vehicle traveling on a road with a simple configuration. The width or height of vehicles V1 to V3 obtained in this manner can be used for collecting statistics on vehicles traveling on the corresponding road, determining the vehicle type, determining whether or not there is a traffic law violation such as a traffic lane violation, and so on.

[0020] Furthermore, according to this embodiment, it is possible to accurately estimate the width and height of vehicles that have similar external shapes but different sizes, such as Japanese light trucks and commercial trucks, and light vans and commercial vans. This is because the present invention employs a configuration that estimates at least one of the width and height of vehicles V1 to V3 based on the size ratio to a subject of known size. On the other hand, with the method using the learning model described in the background art, it is impossible to accurately estimate the width or height of a vehicle unless the distance to the vehicle is accurately measured and the vehicle type is identified. This is also one of the advantages of the present disclosure.

[0021] [First Embodiment] Next, a first embodiment in which the vehicle width is estimated for a specific type of vehicle will be described. Fig. 5 is a diagram showing one configuration of the present disclosure. Referring to Fig. 5, a configuration including a dimension estimation system 100 and a portable camera C connected to the dimension estimation system 100 is shown. Below, an example will be described in which the vehicle width of a specific type of vehicle V is estimated for a cab-over type vehicle such as a truck or bus with a driver's seat located above the engine.

[0022] The dimension estimation system 100 includes a vehicle extraction unit 101, a front end extraction unit 102, a license plate extraction unit 103, a vehicle width estimation unit 104, a recording unit 105, an analysis result database (analysis result DB) 106, and three inference models 111 to 113. The inference models 111 to 113 are also called learning models.

[0023] The vehicle extraction unit 101 uses a first inference model 111 to extract an image of a vehicle V from an image captured by a camera C. The first inference model 111 is learned and tuned using training data prepared in advance, and is generated so that it can detect objects (vehicles) in an image.

[0024] The front end extraction unit 102 uses a second inference model 112 to extract the area (contour) of the front end of the vehicle V from the image captured by the camera C. The second inference model 112 is generated by learning and tuning using pre-prepared training data so that it can detect the front end of an object (vehicle) in an image. As described above, in this embodiment, a cab-over vehicle is the target, and therefore the front end is the part of the front face of the vehicle where the bumper, front mask, front lights, windshield, etc. are arranged on approximately the same plane. This front end extraction unit 102 corresponds to the extraction unit 11 described above.

[0025] The license plate extraction unit 103 uses a third inference model 113 to extract the area (contour) of the license plate portion of the vehicle V from the image captured by the camera C. The third inference model 113 is generated by learning and tuning using pre-prepared training data so that it can detect the license plate portion of an object (vehicle) in the image. Note that a license plate reading function may be added to this license plate extraction unit 103.

[0026] The vehicle width estimation unit 104 estimates the vehicle width of a cab-over vehicle using the areas (outlines) extracted by the vehicle extraction unit 101, the front end extraction unit 102, and the license plate extraction unit 103. This vehicle width estimation unit 104 corresponds to the estimation unit described above.

[0027] The recording unit 105 records the images extracted by the vehicle extraction unit 101 , front end extraction unit 102 and license plate extraction unit 103 and the vehicle width estimated by the vehicle width estimation unit 104 in the analysis result DB 106 .

[0028] In this embodiment, a portable camera C equipped with a tripod is used as the camera C. The advantage of using such a portable camera C is that it can be freely carried to areas where an increase in traffic volume of specific types of vehicles is expected or areas where enforcement of violations is desired. On the other hand, changing the position of the camera C may affect the detection performance of the vehicle extraction unit 101, front end extraction unit 102, and license plate extraction unit 103. The desirable positional relationship between the camera C and the vehicle will be described below.

[0029] FIG. 6 shows the horizontal angle θ of the camera C when the camera is installed on the roadside. ０ (For example, θ ０ = 50°). The position (measurement point) where the vehicle V is photographed is determined in advance. The vehicle V approaches the camera C, and when it reaches the measurement point at a distance 1, the camera C photographs the vehicle V. At this time, the camera C is positioned at an angle θ １ By preparing training data captured under such conditions, the detection performance of the vehicle extraction unit 101, the front end extraction unit 102, and the license plate extraction unit 103 can be maintained.

[0030] FIG. 7 shows the vertical angle θ of the camera C installed at a height t. ２ When the vehicle V is photographed at the timing when the vehicle V arrives at the measurement point, the camera C is positioned at an angle θ ３ The vehicle V is captured by the above equation. More preferably, training data that also matches the height condition is prepared. This allows the detection performance of the vehicle extraction unit 101, the front end extraction unit 102, and the license plate extraction unit 103 to be maintained.

[0031] The installation location of camera C is not limited to the roadside as shown in Figure 6. For example, it is also desirable to select a position on the road where vehicle V can be captured as directly in front as possible, using a pedestrian bridge or T-junction as shown in Figure 8, and prepare a corresponding inference model. This can further improve the detection performance of vehicle extraction unit 101, front end extraction unit 102, and license plate extraction unit 103.

[0032] FIG. 9 shows guide lines for adjusting the angle of view that are superimposed on the display for confirming the image captured by camera C. The dotted lines in FIG. 9 are guide lines GL for adjusting the angle of view. The upper diagram in FIG. 9 shows the state before the angle of view is adjusted, in which the size of the vehicle V is small relative to the guide lines GL for adjusting the angle of view. From this state, by increasing the magnification using the zoom function of camera C, an image suitable for recognizing an object (vehicle), as shown in the lower diagram in FIG. 9, can be obtained. In the example in FIG. 9, an example of enlarging the image of the vehicle using zoom has been described, but by using the guide lines GL in FIG. 9, the position (side view angle) and height (depression angle) of the camera can also be adjusted at the same time.

[0033] FIG. 10 is a diagram illustrating the size of license plates (automobile registration plates). The symbol NPL in FIG. 10 indicates the size of large license plates (large plates) attached to standard freight vehicles with a gross vehicle weight of 8 tons or more or a maximum load capacity of 5 tons or more, and standard passenger vehicles with a passenger capacity of 30 or more in Japan. The symbol NPM indicates the size of medium license plates (medium plates) attached to vehicles other than vehicles with large plates and small two-wheeled vehicles in Japan. As shown in FIG. 10 , there are essentially two license plate sizes for automobiles (excluding light vehicles) in Japan. Furthermore, by narrowing down the vehicle to either a large plate or a medium plate based on vehicle characteristics, the license plate size can also be narrowed down to one. In the following explanation, this will be used to estimate the vehicle width.

[0034] Next, the operation of this embodiment will be described in detail with reference to the drawings. Fig. 7 is a diagram showing the overall operation of one configuration. Referring to Fig. 11, first, the dimension estimation system 100 extracts a vehicle from an image captured by camera C (step S001).

[0035] Next, the dimension estimation system 100 extracts the region of the front end of the vehicle from the image (step S002). For example, the dimension estimation system 100 extracts the region of the front end of the vehicle indicated by the dashed line from the image shown in FIG.

[0036] Next, the dimension estimation system 100 extracts the area of ​​the vehicle license plate from the image (step S003). For example, the dimension estimation system 100 extracts the area of ​​the vehicle license plate indicated by the dashed line from the image shown in FIG.

[0037] Next, the dimension estimation system 100 calculates the vehicle width using each of the extracted regions (step S004). For example, the dimension estimation system 100 calculates the vehicle width by multiplying the value obtained by dividing the width (number of pixels) w2 of the region at the front end of the vehicle shown in Figure 3 by the width (number of pixels) w1 of the license plate by the known width of the license plate.

[0038] Finally, the dimension estimation system 100 records the calculated vehicle width together with the extracted image in the analysis result DB 106 (step S005).

[0039] The vehicle width with the image thus obtained can be used to collect vehicle width data for specific types of vehicles traveling on the road in question, to identify vehicle types, and to determine whether or not there are any violations of traffic laws, such as traffic lane violations. In particular, in this embodiment, cab-over vehicles are targeted for vehicle width estimation. Because these vehicles have a flat surface at the front end, their distinctive front end and license plate can be photographed at approximately the same angle even from an oblique side view. Therefore, the dimension estimation system 100 of this embodiment can accurately estimate the vehicle width of these vehicles.

[0040] Second Embodiment Next, a second embodiment will be described in which a function for checking the consistency of extracted images is added to the dimension estimation system 100a. Fig. 12 is a diagram showing an example configuration of another dimension estimation system 100a according to the present disclosure. The difference from the example configuration shown in Fig. 5 is that a confirmation unit 107 is added to the dimension estimation system 100a. Since the other configurations are the same as those in Fig. 5, the following description will focus on the differences.

[0041] The confirmation unit 107 confirms the consistency between the vehicle detection result by the inference model 111 of the vehicle extraction unit 101 and the detection result of the front end of the vehicle by the inference model 112 of the front end extraction unit 102. For example, if the area on the front end side of the vehicle according to the inference model 112 of the front end extraction unit 102 is not recognized as the front end of the vehicle, the confirmation unit 107 determines that there is no consistency. If it determines that there is no consistency, processing such as discarding the relevant data or extracting the image again is performed.

[0042] Fig. 13 is a flowchart showing the operation of the dimension estimation system 100a. The difference from the flowchart shown in Fig. 11 is that after extracting the license plate portion in step S003, the dimension estimation system 100a checks the consistency of the extracted area (step S010).

[0043] FIG. 14 is a diagram illustrating the process of checking the consistency between extracted regions in step S010. In the left diagram of FIG. 14, a bounding box BB2 indicating the region from which the image of the vehicle's front end is extracted is positioned appropriately relative to a bounding box BB1 indicating the region from which the image of the vehicle is extracted. In this case, it is determined that the extracted regions are consistent. On the other hand, in the right diagram of FIG. 14, a bounding box BB3 indicating the region from which the image of the vehicle's front end is extracted is outside the bounding box BB1, significantly shifted to the left. In this case, the dimension estimation system 100a determines that the extracted regions are inconsistent. This prevents the vehicle width estimation unit 104 from estimating the vehicle width based on a region wider or shifted than the actual vehicle's front end.

[0044] As described above, according to this embodiment in which a confirmation unit is added, it is possible to improve the accuracy of estimating the vehicle width. While the above embodiment shows an example in which a bounding box (rectangle) is set to recognize an object, other methods may be used. For example, a method may be employed in which an area is designated by tracing the outline (contour) of the vehicle, and the object in the designated area is recognized.

[0045] In the above embodiment, the confirmation unit 107 is described as confirming the consistency between the vehicle detection result and the detection result of the front end of the vehicle, but the license plate may also be subject to the consistency confirmation. In this case, a configuration can be adopted in which the confirmation unit 107 confirms the consistency between the vehicle detection result, the detection result of the front end of the vehicle, and the detection result of the license plate. Furthermore, a configuration can be adopted in which a second confirmation unit is added, separate from the confirmation unit 107, to confirm the consistency between the license plate detection result and the vehicle detection result.

[0046] [Third Embodiment] Next, a third embodiment will be described in which a function for reporting violating vehicles using estimated vehicle widths is added to the dimension estimation system 100b. Fig. 15 is a diagram showing an example configuration of another dimension estimation system 100b according to the present disclosure. The difference from the example configuration shown in Fig. 12 is that a determination unit 120 and a determination result output device 121 are added to the dimension estimation system 100b. Since the other configurations are the same as those of Fig. 12, the following description will focus on the differences.

[0047] The judgment unit 120 reads out the new entry added to the analysis result DB 106, judges whether the vehicle width satisfies the conditions set forth in traffic regulations, etc., and drives the judgment result output device 121 based on the result.

[0048] The determination result output device 121 is equipped with a red light R and a green light G. When the determination unit 120 determines, as a result of the determination, that the width of the newly detected vehicle does not satisfy the conditions stipulated in traffic laws and regulations, etc., it turns on the red light R of the determination result output device 121. On the other hand, when the width of the newly detected vehicle satisfies the conditions stipulated in traffic laws and regulations, it turns on the green light G of the determination result output device 121.

[0049] 16 is a flow chart showing the operation of the dimension estimation system 100b. Referring to FIG. 16, the dimension estimation system 100b first reads new data from the analysis result DB 106 (step S201).

[0050] Next, the dimension estimation system 100b compares the vehicle width value in the read data with the conditions stipulated in traffic regulations and the like to determine whether the vehicle is in violation of the traffic regulations (step S202).

[0051] If the result of the determination is that the vehicle is a violating vehicle (Yes in step S202), the dimension estimation system 100b turns on the red light R of the determination result output device 121 (step S203). On the other hand, if the result of the determination is that the vehicle is not a violating vehicle (No in step S202), the dimension estimation system 100b turns on the green light G of the determination result output device 121. The dimension estimation system 100b turns on the green light G of the determination result output device 121 (step S204).

[0052] As described above, with the configuration including the determination unit 120 and the determination result output device 121, the dimension estimation system 100b can be used for enforcing traffic laws and regulations.

[0053] In the above description, the judgment result output device 121 equipped with a red light R and a green light G is used, but the device for outputting the judgment result is not limited to this. For example, a judgment result output device that can output the judgment result not only by a red light but also by audio, buzzer, or image output may be used. Furthermore, the judgment result does not need to be a binary value of whether or not a violation has occurred, and may be output as true or false, or according to the degree of violation.

[0054] [Fourth Embodiment] Next, a fourth embodiment will be described in which a function for matching with special vehicle permit application data is added to the dimension estimation system 100c. Figure 17 is a diagram showing an example configuration of another dimension estimation system 100c of the present disclosure. The difference from the example configuration shown in Figure 12 is that a determination unit 120a and a pre-application information database (pre-application information DB) 130 have been added to the dimension estimation system 100c. Since the other configurations are the same as those of Figure 12, the following description will focus on the differences.

[0055] The advance application information DB 130 stores information about vehicles that have applied in advance to travel on roads where cameras C are installed. For example, the advance application information DB 130 stores applicant information such as the applicant's name and address, and application details such as the start date of travel, vehicle type classification, and license plate information.

[0056] The determination unit 120a reads data from the analysis result DB 106, and if the vehicle width exceeds a predetermined standard, determines whether there is corresponding data in the advance application information DB 130. Note that, in order to compare with the advance application information DB 130, license plate information must be registered in the data of the analysis result DB 106. This license plate information may be read by the license plate extraction unit 103, or may be read from the license plate image in the analysis result DB 106 using a separate license plate reader.

[0057] If the determination unit 120a determines that a prior application has not been made for a vehicle whose width exceeds the predetermined standard, the determination unit 120a records that there has been a vehicle width violation in the corresponding data in the analysis result DB 106. On the other hand, if a prior application has been made for a vehicle whose width exceeds the predetermined standard, the determination unit 120a does not determine that there has been a vehicle width violation.

[0058] 18 is a flowchart showing the operation of the dimension estimation system 100c. Referring to Fig. 18, the dimension estimation system 100c first extracts one piece of data from the analysis result DB 106 and compares it with data in the advance application information DB (step S301).

[0059] If the result of the comparison shows that the violating vehicle is not registered in the advance application information DB 170 (Yes in step S302), the dimension estimation system 100c determines that the vehicle is a violating vehicle for which no advance application has been made, and records the corresponding vehicle in the analysis result DB 106 as a violating vehicle (step S303). On the other hand, if the violating vehicle is registered in the advance application information DB 170 (No in step S302), the dimension estimation system 100c determines that the corresponding vehicle in the analysis result DB 106 is not a violating vehicle.

[0060] Next, the dimension estimation system 100c determines whether matching has been completed for all data in the analysis result DB 106 (step S304). If matching has not been completed for all data (No in step S304), the dimension estimation system 100c reads the next data and repeats the processes in steps S301 to S303. If matching has been completed for all data (Yes in step S304), the dimension estimation system 100c ends the process (Yes in step S304).

[0061] As described above, it is possible to identify violating vehicles without prior application, according to a configuration including the determination unit 120a and the determination result output device 121. In particular, with this configuration, the matching process shown in Fig. 18 can be performed as a batch process independently of the vehicle extraction process and the species width estimation process, which has the advantage that it can be performed even on a computer with relatively low performance.

[0062] [Fifth Embodiment] In the above-described embodiments, a portable camera C is used as the camera C. However, if the measurement area is determined in advance, a fixed camera may be installed, or an existing camera for photographing the road or a moving vehicle may be used. Figure 19 is a diagram showing an example configuration of another dimension estimation system 100d according to the present disclosure. The difference from the example configuration shown in Figure 5 is that the dimension estimation system 100d uses a fixed camera Cd attached to a traffic light S, rather than a portable camera C. The other configurations are similar to those of Figure 5, etc., and therefore will not be described again.

[0063] As with the first to fourth embodiments, the configuration shown in Fig. 19 also makes it possible to estimate the width and height of a vehicle traveling on a road with a simple configuration. Note that, although the example in Fig. 19 uses a fixed camera Cd attached to a traffic light S, the example of the fixed camera is not limited to this. For example, a fixed camera for traffic control or a fixed camera for monitoring traffic congestion can also be used as the fixed camera.

[0064] Although the embodiments of the present disclosure have been described above, the present disclosure is not limited to the above-described embodiments, and further modifications, substitutions, and adjustments can be made without departing from the basic technical concept of the present disclosure. For example, the network configurations, element configurations, and data representation formats shown in the drawings are examples intended to aid in understanding the present disclosure, and are not limited to the configurations shown in these drawings.

[0065] For example, while the first to fifth embodiments have been described as estimating the vehicle width, a similar method can also be used to calculate the vehicle height. Furthermore, while the first to fifth embodiments have been described as estimating the vehicle width of a cab-over vehicle, other vehicle types may also be used as the vehicle width estimation target. For example, the dimension estimation systems 100 to 100d may estimate the vehicle width of a bonnet-type vehicle. In this case, too, the dimension estimation systems 100 to 100d extract the front end and estimate the vehicle width. In the case of a bonnet-type vehicle, the dimension estimation systems 100 to 100d may be configured to extract an area including the front bumper, front grille, headlights, bonnet, etc. as the front end. Furthermore, in addition to the above, the dimension estimation systems 100 to 100d may extract an area including the windshield as the front end. Furthermore, all vehicles captured by the cameras C and Cd may be used as the vehicle width estimation target.

[0066] Furthermore, in the first to fifth embodiments described above, a license plate is used as the "subject of known size," but other subjects can also be used as the "subject of known size." For example, if there is a sign or the like that is required to be attached to a specific type of vehicle, the sign can also be used as the "subject of known size."

[0067] (Hardware Configuration) In each embodiment of the present disclosure, each component of each device represents a functional unit block. Some or all of the components of each device are realized by an arbitrary combination of an information processing device 900 and a program, for example, as shown in FIG. 20 . FIG. 20 is a block diagram showing an example of the hardware configuration of the information processing device 900 that realizes each component of each device. The information processing device 900 includes, as an example, the following configuration: - CPU (Central Processing Unit) 901 - ROM (Read Only Memory) 902 - RAM (Random Access Memory) 903 - Program 904 loaded into RAM 903 - Storage device 905 that stores the program 904 - Drive device 907 that reads and writes to a recording medium 906 - Communication interface 908 that connects to a communication network 909 - Input / output interface 910 that inputs and outputs data - Bus 911 that connects each component

[0068] Each component of each device in each embodiment is realized by the CPU 901 acquiring and executing a program 904 that realizes the function. That is, the CPU 901 in FIG. 20 executes a vehicle extraction program and a vehicle width estimation program, and performs an update process for each calculation parameter stored in the RAM 903, the storage device 905, etc. The program 904 that realizes the function of each component of each device is stored in the storage device 905 or the ROM 902 in advance, for example, and is read out by the CPU 901 as needed. The program 904 may be supplied to the CPU 901 via the communication network 909, or may be stored in advance on the recording medium 906, and the drive device 907 may read out the program and supply it to the CPU 901.

[0069] There are various variations in the implementation of each device. For example, each device may be implemented by any combination of a separate information processing device 900 and a program for each component. Furthermore, multiple components of each device may be implemented by any combination of a single information processing device 900 and a program. That is, each unit (processing means, function) of the dimension estimation system shown in the first to fifth embodiments can be implemented by a computer program that causes a processor installed in the device to execute the above-described processes using its hardware. Similarly, each unit (processing means, function) of the passable width estimation system shown in the fifth embodiment can be implemented by a computer program that causes a processor installed in the device to execute the above-described processes using its hardware.

[0070] In addition, some or all of the components of each device may be realized by other general-purpose or dedicated circuits, processors, etc., or a combination of these. These may be configured by a single chip, or by multiple chips connected via a bus.

[0071] Some or all of the components of each device may be realized by a combination of the above-mentioned circuits and programs.

[0072] When some or all of the components of each device are realized by multiple information processing devices, circuits, etc., the multiple information processing devices, circuits, etc. may be centrally or decentralized. For example, the information processing devices, circuits, etc. may be realized as a client-server system, a cloud computing system, or the like, in a form in which each device is connected via a communication network.

[0073] It should be noted that the above-described embodiments are preferred embodiments of the present disclosure, and the scope of the present disclosure is not limited to only the above-described embodiments. In other words, those skilled in the art can modify or substitute the above-described embodiments to construct various modified forms without departing from the gist of the present disclosure.

[0074] A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes.

[0075] [Supplementary Note 1] A vehicle dimension estimation system comprising: an extraction unit connectable to a camera capable of photographing vehicles traveling on a road, and extracting a measured area of ​​the vehicle from an image photographed by the camera; and an estimation unit estimating at least one of the vehicle width or height based on a size ratio between the extracted measured area of ​​the vehicle and a pre-set object appearing in the image. [Supplementary Note 2] In the above dimension estimation system, the object may be the license plate of the vehicle. [Supplementary Note 3] In the above dimension estimation system, the object may be a lane of a road on which the vehicle is traveling. [Supplementary Note 4] In the above dimension estimation system, the extraction unit may be configured to extract the vehicle and the measured area using a first inference model for detecting the vehicle and a second inference model for extracting the measured area of ​​the vehicle, respectively. [Supplementary Note 5] In the above-mentioned dimension estimation system, the extraction unit may further extract the subject using a third inference model for detecting the subject. [Supplementary Note 6] In the above-mentioned dimension estimation system, the extraction unit may further include a confirmation unit that confirms consistency between the detection results using the first inference model and the detection results using the second inference model. [Supplementary Note 7] In the above-mentioned dimension estimation system, the extraction unit may further include a confirmation unit that confirms consistency between the detection results using the first inference model, the detection results using the second inference model, and the detection results using the third inference model. [Supplementary Note 8] In the above-mentioned dimension estimation system, the extraction unit detects the specific model of vehicle, the estimation unit estimates at least one of the width or height of the specific model of vehicle, and the system may further include a determination unit that determines whether the specific model of vehicle satisfies predetermined conditions, and the system may function as a non-compliant vehicle detection device that detects vehicles that do not satisfy the conditions. [Supplementary Note 9] The extraction unit of the dimension estimation system described above may be configured to extract the front end of a cab-over vehicle or a bonnet-type vehicle as the measurement region of the vehicle.[Supplementary Note 10] A method for estimating the front dimension of a vehicle, comprising: extracting a measurement area of ​​the vehicle from an image captured by a camera capable of capturing images of the vehicle traveling on a road; and estimating at least one of the vehicle width and vehicle height based on the size ratio between the extracted measurement area of ​​the vehicle and a pre-set subject appearing in the image. The embodiments described in each of the above Supplements can be combined with each other after making necessary modifications. For example, a configuration that combines the contents of Supplementary Note 2 and Supplementary Note 3 is also included within the scope of disclosure of this specification. The embodiment of Supplementary Note 10 can be expanded into the embodiments of Supplements 2 to 9, similar to Supplementary Note 1.

[0076] The disclosures of the above-cited patent documents are incorporated herein by reference and may be used as the basis or part of this disclosure, as necessary. Modifications and adjustments of the embodiments and examples are possible within the scope of this disclosure (including the claims), and further based on its basic technical concept. Furthermore, various combinations and selections (including partial deletions) of various disclosed elements (including elements of each claim, each element of each embodiment or example, each element of each drawing, etc.) are possible within the scope of this disclosure. In other words, this disclosure naturally includes various modifications and alterations that would be possible by a person skilled in the art in accordance with the entire disclosure, including the claims, and the technical concept. In particular, with regard to the numerical ranges described herein, any numerical value or subrange within that range should be construed as specifically described, even if not otherwise specified. Furthermore, the disclosures of the above-cited documents, when used in part or in whole in combination with the disclosures herein as part of this disclosure, in accordance with the spirit of this disclosure, are also deemed to be included in the disclosures of this application.

[0077] 10, 100, 100a, 100b, 100c Dimension estimation system 11 Extraction unit 12 Estimation unit C, Cd Camera 101 Vehicle extraction unit 102 Front end extraction unit 103 License plate extraction unit 104 Vehicle width estimation unit 105 Recording unit 106 Analysis result database (analysis result DB) 107 Confirmation unit 111-113 Inference model 120, 120a Determination unit 121 Determination result output device 130 Advance application information database (advance application information DB) 900 Information processing device 901 CPU (Central Processing Unit) 902 ROM (Read Only Memory) 903 RAM (Random Access Memory) 904 Program 905 Storage device 906 Recording medium 907 Drive device 908 Communication interface 909 Communication network 910 Input / output interface 911 Bus NPL Number plate (large plate) NPM Number plate (medium plate) BB1, BB2 Bounding box R Red light G Green light V, V1 to V3 Vehicle w1 Width of number plate w2 Vehicle width w3 Width of lane θ ０ Horizontal angle of view of camera C θ １ , θ ３ Shooting angle θ ２ Vertical angle of view of camera C

Claims

1. A vehicle dimension estimation system comprising: an extraction unit that can be connected to a camera that can photograph vehicles traveling on a road, and that extracts a measured area of ​​the vehicle from an image photographed by the camera; and an estimation unit that estimates at least one of the vehicle width and height of the vehicle based on a size ratio between the extracted measured area of ​​the vehicle and a previously set subject that appears in the image.

2. The vehicle size estimation system of claim 1, wherein the subject is a license plate of the vehicle.

3. The vehicle size estimation system according to claim 1, wherein the subject is a lane of a road on which the vehicle is traveling.

4. A vehicle dimension estimation system according to any one of claims 1 to 3, wherein the extraction unit extracts the vehicle and the measured area using a first inference model for detecting the vehicle and a second inference model for extracting the measured area of ​​the vehicle.

5. The vehicle dimension estimation system of claim 4, wherein the extraction unit further extracts the object using a third inference model for detecting the object.

6. A vehicle dimension estimation system as claimed in claim 4, further comprising a verification unit for verifying consistency between the detection results obtained by said first inference model and the detection results obtained by said second inference model.

7. A vehicle dimension estimation system as claimed in claim 5, further comprising a verification unit that verifies the consistency between the detection results based on the first inference model, the detection results based on the second inference model, and the detection results based on the third inference model.

8. A dimension estimation system according to any one of claims 1 to 7, wherein the extraction unit detects the specific type of vehicle, the estimation unit estimates at least one of the vehicle width or height of the specific type of vehicle, and the system further includes a judgment unit that judges whether the specific type of vehicle satisfies pre-set conditions, and functions as a violating vehicle detection device that detects vehicles that do not satisfy the conditions.

9. A vehicle dimension estimation system according to any one of claims 1 to 8, wherein the extraction unit extracts a front end portion of a cab-over type vehicle or a bonnet type vehicle as the measured area of ​​the vehicle.

10. A method for estimating vehicle dimensions, comprising: extracting a measurement area of ​​a vehicle traveling on a road from an image captured by a camera capable of capturing images of the vehicle; and estimating at least one of the vehicle width and height based on a size ratio between the extracted measurement area of ​​the vehicle and a previously set subject appearing in the image.

Images