JP2017045137A - Vehicle type discrimination device and vehicle type discrimination method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve accuracy of vehicle type discrimination.SOLUTION: A vehicle type discrimination device includes: a storage part for storing first shape information showing shape of a loading space or a load or a passenger car existing at the rear part of a vehicle; an acquisition part for acquiring an image of the vehicle picked up by an imaging part; and a vehicle type discrimination part for discriminating the type of the vehicle from the loading space or the load or the passenger car existing at the rear part of the vehicle by checking with the rear area of the vehicle shown in the image acquired from the acquisition part and the first shape information stored in the storage part.SELECTED DRAWING: Figure 3

Description

  The present application relates to a vehicle type identification device and a vehicle type identification method.

  In order to protect road exhaustion, etc., the types of vehicles subject to control in the Road Traffic Law are subdivided. For example, for large vehicles such as tank trucks, the width, length, height, weight, etc. Regulatory conditions are established.

JP 2014-164601 A Japanese Patent Laid-Open No. 2002-008188 JP 2006-012178 A

  However, in the prior art, vehicle type discrimination such as a large vehicle for automating the above-described control is insufficient, and high-precision vehicle type discrimination is required.

  Therefore, a vehicle type discrimination device and a vehicle type discrimination method that improve the accuracy of vehicle type discrimination are provided.

  The vehicle type identification device according to the embodiment acquires a vehicle image captured from an imaging unit, a storage unit that stores first shape information that represents the shape of a loading platform, a luggage, or a passenger car that exists behind the vehicle. The vehicle, a rear region of the vehicle represented in the image obtained from the acquisition unit, and the first shape information stored in the storage unit, and A vehicle type discriminating unit that discriminates the vehicle type of the vehicle from an existing loading platform, luggage, or passenger car.

FIG. 1 is a diagram illustrating an example of a schematic configuration of a vehicle type identification system according to the first embodiment. FIG. 2 is a diagram illustrating an example of a camera image obtained from a monocular camera. FIG. 3 is a diagram illustrating an example of a functional configuration of the vehicle type identification device according to the first embodiment. FIG. 4 is a diagram illustrating a dictionary pattern for vehicle detection according to the first embodiment. FIG. 5 is a diagram illustrating the trailer component DB according to the first embodiment. FIG. 6 is a flowchart illustrating an example of a vehicle type determination process according to the first embodiment. FIG. 7 is a diagram illustrating a large vehicle to be authenticated. FIG. 8 is a diagram illustrating an image generated by deleting the tractor unit. FIG. 9 is a diagram illustrating an example of a schematic configuration of a vehicle type identification system according to the second embodiment. FIG. 10 is a diagram illustrating an example of a functional configuration of the vehicle type identification device according to the second embodiment. FIG. 11 is a diagram illustrating a group of three-dimensional crane parts according to the second embodiment. FIG. 12 is a diagram illustrating a partial image DB viewed from the side of the vehicle according to the second embodiment. FIG. 13 is a diagram illustrating a tow bar part group having a three-dimensional shape according to the second embodiment. FIG. 14 is a flowchart illustrating an example of a vehicle type determination process according to the second embodiment. FIG. 15 is a diagram illustrating an example of detecting a crane portion. FIG. 16 is a diagram illustrating an example of space measurement at the rear of the tractor. FIG. 17 is a diagram illustrating an example of detection of a tow bar.

(First embodiment)
The vehicle type discrimination system in the first embodiment will be described.

(Schematic configuration example of vehicle type identification system)
FIG. 1 is a diagram illustrating an example of a schematic configuration of a vehicle type identification system according to the first embodiment. A vehicle type identification system 1 shown in FIG. 1 includes a monocular camera (imaging unit) 10, a vehicle detector 20, a vehicle type identification device 30 that is an example of an electronic computer (information processing device), and a vehicle weight scale 40. The monocular camera 10, the vehicle detector 20, the vehicle type identification device 30, and the vehicle weight scale 40 are connected in a state where data can be transmitted and received through a wired or wireless communication network 50.

  The vehicle type discrimination system 1 shown in the example of FIG. 1 is installed on, for example, a road 70 through which a large vehicle 60 passes, a parking lot entrance / exit, a toll gate on a highway, and the like, but is not limited thereto.

  The monocular camera 10 is installed so as to photograph an obliquely upper part of the vehicle 60, but the installation position is not limited to this. The monocular camera 10 is adjusted so as to look down at a height of about 5 to 7 m from the upper side of the vehicle 60 (the legal height of the vehicle height is 3.8 m) and a depression angle of about 25 to 45 degrees. When the monocular camera 10 is installed, the angle of view is set for the vehicle 60 moving within the field of view so that the entire view of the vehicle 60 is reflected.

  The entire view of the imaging target vehicle 60 is transferred to the vehicle type identification device 30 when the vehicle 60 moves to a predetermined position with respect to the camera angle of view of the monocular camera 10.

  The vehicle detector 20 is, for example, a laser sensor (infrared sensor or the like) installed on the road 70, but is not limited thereto. For example, the vehicle detector 20 may detect the approach of the vehicle by scanning a camera image from the monocular camera 10. The vehicle detector 20 generates a shooting trigger for shooting with the monocular camera 10 when the vehicle 60 enters, and transmits the shooting trigger to the monocular camera 10.

  The vehicle type discriminating device 30 discriminates the vehicle type information using, for example, an internal algorithm (a vehicle type discrimination program or the like). The vehicle type identification device 30 acquires the camera image obtained from the monocular camera 10 and the vehicle weight obtained from the vehicle weight meter 40 at the timing of the detection result obtained from the vehicle detector 20, for example, and based on the obtained various information. Car model discrimination is performed.

  The vehicle type in the present embodiment is information for distinguishing vehicles. Examples of vehicle types include large vehicles, passenger cars, and motorcycles, but are not limited thereto. The vehicle type includes information on the shape of the vehicle (for example, information on one or more shapes of the width, height, and length of the vehicle) in addition to the classification of the vehicle.

  The vehicle weight scale 40 may be embedded in the road 70, may be a flat surface such as a plate or iron plate installed on the road 70, or may have a groove or the like on which a tire is placed. The vehicle weight scale 40 measures the weight of the vehicle 60 when the tire of the vehicle 60 passes over the vehicle weight scale 40.

  FIG. 2 is a diagram illustrating an example of a camera image obtained from a monocular camera. As shown in FIG. 2, the angle of view of the monocular camera 10 is, for example, that the lower end of the tractor portion on the front surface of the vehicle 60 and the rear end of the trailer portion existing behind the tractor portion of the vehicle 60 are camera images 80. It is set to fit in the vertical direction. For example, in the case of a large vehicle, the vehicle 60 has a long overall length. Therefore, the monocular camera 10 may be photographed, for example, tilted by 90 degrees in the roll direction so that the vehicle 60 shown in FIG.

  From the camera image 80 shown in FIG. 2, the vehicle type identification device 30 according to the first embodiment includes a tractor portion (front region) 61 having a predetermined shape including a driver's seat of the vehicle 60, and a rear stage (loading platform) of the vehicle 60. The trailer portion (rear region) 62 is identified, and the rear region 62 is used for vehicle type discrimination. In the first embodiment, by using the rear region 62, which is a characteristic part, for the vehicle type determination instead of the front region 61 of a large vehicle that often has a common shape, more accurate vehicle type determination can be realized. It becomes possible.

(Functional configuration example of the vehicle type identification device 30)
Next, a functional configuration example of the above-described vehicle type identification device 30 will be described. FIG. 3 is a diagram illustrating an example of a functional configuration of the vehicle type identification device according to the first embodiment. The vehicle type identification device 30 shown in the example of FIG. 3 includes a storage unit 90, an acquisition unit 91, a vehicle detection unit 92, a tractor unit detection unit 93, a trailer unit cutout unit (cutout unit) 94, and a matching unit. 95, a vehicle type determination unit 96, a network connection unit 97, and a control unit 98.

  By the way, when a large vehicle travels on a road, it often places a burden on the road as compared to a normal vehicle. For this reason, problems such as unauthorized operation of special vehicles included in large vehicles and overloading operations occur. Along with this, in recent years, it has been required to accurately determine the type of a large vehicle.

  Therefore, the vehicle type identification device 30 of the first embodiment determines the vehicle type in detail for large vehicles.

  In the first embodiment, the large vehicle is configured to include a tractor portion and a trailer portion. The tractor portion is a front portion (leading portion, towing vehicle) including a loading platform and a driver seat provided for towing a passenger car among large vehicles. The trailer part is a rear part (following part, towed vehicle) towed by a tractor part, including a cargo bed, luggage, and passenger car among large vehicles. Note that the luggage placed on the cargo bed may also be included in the trailer section. A large vehicle including the towed vehicle described above will be described as a towing vehicle.

  The storage unit 90 stores various types of information necessary for the first embodiment. For example, the storage unit 90 stores various programs (applications, software, etc.) for executing vehicle type determination processing, various setting information, and the like.

  The storage unit 90 stores a dictionary pattern 101 for vehicle detection. FIG. 4 is a diagram illustrating a dictionary pattern for vehicle detection according to the first embodiment. As shown in FIG. 4, the vehicle detection dictionary pattern 101 is a database in which a shape pattern (second shape information) indicating the shape of the tractor portion of the vehicle is stored.

  In addition, the storage unit 90 stores a trailer component database (DB) 201. The trailer component DB 201 is a database that stores shape patterns (first shape information) indicating basic shapes such as a loading platform, a luggage, and a passenger car (a group of components such as a container) included in the trailer portion of the vehicle. FIG. 5 is a diagram illustrating the trailer component DB according to the first embodiment. As shown in FIG. 5, the trailer component DB 201 stores shape patterns representing vehicle loading platforms, luggage, and passenger cars according to vehicle types. This makes it possible to specify the vehicle type when specifying the shape of the trailer part.

  The information stored in the storage unit 90 may be acquired from another device (for example, a database server) connected via a communication network typified by the Internet or a LAN (Local Area Network). .

  The acquisition unit 91 acquires information obtained from each of the monocular camera 10, the vehicle detector 20, and the vehicle weight meter 40. The acquisition unit 91 may acquire the image of the monocular camera 10 and the vehicle weight information of the vehicle weight meter 40 at the timing when the vehicle detector 20 detects the vehicle 60.

  The acquisition unit 91 acquires an image in which the entire view of the vehicle is captured, including the front surface of the vehicle imaged from the monocular camera 10. In the present embodiment, an example in which a full view of the vehicle including the front surface of the vehicle is captured will be described. However, the acquisition unit 91 may acquire image data including the rear surface of the vehicle.

  The vehicle detection unit 92 detects the vehicle 60 from the camera image obtained from the monocular camera 10. Specifically, whether the vehicle is a vehicle, the size of the vehicle (for example, a large vehicle, a small vehicle, a motorcycle) or the like is detected from the rough size of the object included in the image.

  When the vehicle 60 is a large vehicle, the tractor part detection part 93 detects the front area 61 which is a tractor part of the vehicle 60 as shown in FIG.

  The tractor unit detection unit 93 detects the front region 61 by performing matching with the shape pattern stored in the vehicle detection dictionary pattern 101 of the storage unit 90. In the vehicle detection dictionary pattern 101, it is preferable to register a pattern (virtual model) in which the appearance of the tractor portion is estimated based on the approach speed of the vehicle, the installation parameters of the monocular camera 10, and the like.

  The trailer section cutout section 94 cuts out a trailer section to be identified. For example, the trailer section cutout section 94 separates the front area 61 detected by the tractor section detection section 93 from the camera image acquired from the monocular camera 10, the background reflected in the camera image, and the surrounding road portion. A rear region 62 that is a trailer portion is cut out.

  The matching unit 95 uses the rear region 62 obtained by the trailer unit cutting unit 94 as a gaze region, such as a container group or the like pre-stored in the trailer component DB 201 of the storage unit 90 (cargo (package) ) (Matching) with a shape pattern indicating a basic shape to be classified.

  In addition, there is a composite configuration in which a plurality of parts are combined, such as a vehicle transport container, in a vehicle type to be classified. Therefore, the matching unit 95 may calculate a matching score by performing matching on each component group that allows a combination of a plurality of components.

  The vehicle type determination unit 96 determines the vehicle type based on the matching score obtained from the matching unit 95. In addition, the vehicle type determination unit 96 may determine the vehicle type based on the matching score added (or accumulated) by the combination of the parts. The vehicle type discriminating unit 96 acquires shape information such as the width, length, and height corresponding to the discriminated vehicle type based on the above-described dictionary pattern for vehicle detection and the shape pattern indicating the basic shape. It is also possible.

  The network connection unit 97 transmits and receives data with the monocular camera 10, the vehicle detector 20, and the vehicle weight meter 40 via the communication network 50. The network connection unit 97 may be connected to a management server or the like via the Internet, a LAN, or the like, and transmits the vehicle type discrimination result obtained by the vehicle type discrimination device 30 to the management server or the matching described above from the management server. It is also possible to acquire a pattern or the like used in.

  The control unit 98 controls the entire components of the vehicle type identification device 30. The control unit 98 may perform control such as start and end of various processes in the first embodiment, control when an error occurs, and the like.

  In the first embodiment, the vehicle type determination unit 96 may determine whether the vehicle weight obtained from the acquisition unit 91 is equal to or less than the limit weight of the vehicle type determined by the vehicle type determination. In this case, when the vehicle weight acquired from the acquisition unit 91 exceeds the limit weight of the vehicle type, it is possible to notify the management server or the like as a vehicle that violates the regulations.

(Example of vehicle type discrimination process in the first embodiment)
Next, an example of the vehicle type determination process in the first embodiment will be described. FIG. 6 is a flowchart illustrating an example of a vehicle type determination process according to the first embodiment.

  As shown in FIG. 6, the vehicle detection unit 92 of the vehicle type identification device 30 detects a vehicle 60 that is a large vehicle from a camera image obtained from the monocular camera 10 (S10).

  Next, the tractor part detection part 93 detects the tractor part (front area) 61 of the vehicle 60 from the camera image image | photographed with the monocular camera 10 with reference to the dictionary pattern 101 for vehicle detection (S11).

  Next, the trailer section cutout section 94 cuts out the trailer section (rear area) 62 from the camera image acquired by the monocular camera 10 (S12).

  Next, the matching unit 95 matches the trailer unit 62 cut out in the process of S12 with a preset group of parts such as a container (a shape pattern indicating a basic shape) as a recognition target ( S13). In the process of S13, a matching score is calculated.

  Next, the vehicle type discriminating unit 96 discriminates the vehicle type based on the matching score obtained in the process of S13 (S14) and ends the process. For example, in the case of a container for transporting vehicles, since it has a composite part configuration in which a passenger car is mounted on a bridge-like frame, the matching score of both the bridge-like frame and the passenger car becomes relatively large.

  Therefore, in the process of S14, in addition to comparing the individual matching scores of the bridge frame and the passenger car, if the combination of both is possible, the respective matching scores are added and compared as composite part candidates. Add to subject. As a result, the vehicle type can be determined with higher accuracy.

  In the above-described processes of S11 and S13, the accuracy may be improved by increasing the learning patterns to be matched. In addition, in the process of S14, it is possible to improve the vehicle type discrimination accuracy by teaching a sample of a trailer section cut out from a large vehicle to be recognized and newly adding it to a learning sample.

(Specific example of collation with dictionary pattern and trailer section extraction)
Next, a specific example of collation with the dictionary pattern and extraction of the trailer part in the processes of S11 and S12 described above will be described.

  For example, the tractor part detection unit 93 selects the shape pattern 102 that best matches the tractor part of the vehicle 60 with reference to the vehicle detection dictionary pattern 101 shown in FIG.

  As described above, the lower end position of the tractor unit 61 of the vehicle 60 that appears on the screen of the monocular camera 10, the way the tractor unit 61 looks when the vehicle 60 enters, the installation parameters of the monocular camera 10, the entry speed of the vehicle 60, etc. It depends on. Therefore, a shape pattern (virtual model) assuming the above-described change is stored in the vehicle detection dictionary pattern 101 in advance.

  FIG. 7 is a diagram illustrating a large vehicle to be authenticated. In the example shown in FIG. 7, the tractor unit detection unit 93 determines that the shape pattern 102 stored in the vehicle detection dictionary pattern 101 most closely matches the front area 61 included in the vehicle 60. To do. Thereby, the tractor part detection part 93 detects the front area | region 61 included in the vehicle 60 as a tractor part (henceforth "the front area | region 61" is also called "the tractor part 61").

  The trailer section cutout section 94 applies the shape pattern 102 selected as the tractor section 61 to the vehicle 60 shown in FIG. FIG. 8 is a diagram illustrating an image generated by deleting the tractor unit. In the trailer section cutout section 94, the tractor section is erased and the trailer section 62 is represented in the image data 81 shown in FIG.

  Therefore, the trailer section cutout section 94 cuts out the image of the trailer section 62. If the trailer section cutout section 94 is intended to match the tractor section of the vehicle 60 in more detail, the tractor pattern that frequently appears is registered in advance, and is determined by the lower end position of the tractor section and the angle of the monocular camera 10. The boundary with the trailer unit may be obtained by geometric calculation.

(Specific example of vehicle type discrimination by matching with shape pattern)
Next, a specific example of the vehicle type discrimination by collation with the shape pattern in the processes of S13 and S14 described above will be described.

  The matching unit 95 refers to the trailer part DB 201 as shown in FIG. 5 and performs matching with the trailer unit 62 as shown in FIG. 8 cut out by the trailer cutout unit 94. As described above, the trailer component DB 201 stores a component group such as a container (a shape pattern indicating a basic shape).

  In the trailer portion of the vehicle 60, there is a composite component configuration in which a passenger car is combined with a bridge-like frame like a vehicle carrying container. Therefore, it is preferable to prepare shape patterns such as the bridge-like frame 202 and the passenger car 203 in the trailer component DB 201.

  In the case of the vehicle transport container described above, the matching scores of the shape patterns of both the bridge-like frame 202 and the passenger car 203 are relatively large. Therefore, the vehicle type discriminating unit 96 compares the matching score with the shape pattern of the bridge-like frame 202 or the passenger car 203, and compares the matching scores obtained by adding each as a composite part candidate when the combination of both is possible. It is good to target.

  The vehicle type discrimination device 30 according to the first embodiment described above separates the tractor part and the trailer part from the camera image acquired from the monocular camera 10 and discriminates the type of the large vehicle based on the shape of the trailer part. Thereby, it becomes possible to improve the accuracy of vehicle type discrimination of large vehicles.

(Second Embodiment)
Next, a vehicle type identification system according to the second embodiment will be described. In the following description, the same components as those of the vehicle type identification system 1 in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted here.

(Schematic configuration example of vehicle type identification system)
FIG. 9 is a diagram illustrating an example of a schematic configuration of a vehicle type identification system according to the second embodiment. The vehicle type identification system 2 shown in FIG. 9 includes a stereo camera 11, a vehicle type identification device 31 that is an example of an electronic computer (information processing device), and a vehicle weight scale 40. The stereo camera 11, the vehicle type identification device 31, and the vehicle weight scale 40 are connected in a state where data can be transmitted and received through a wired or wireless communication network 50.

  In the second embodiment, the traveling vehicle 60 is photographed from the lateral direction to determine the vehicle. The stereo camera 11 has a plurality of imaging units (for example, a camera 11A and a camera 11B). The stereo camera 11 is provided at a position where the side surface of the vehicle 60 can be photographed, and is adjusted so as to see a field of view with a height of about 2 to 3 m and a road width of about 3.5 m. The content is not limited to this.

  Since the stereo camera 11 employs a stereo system, for example, the camera 11A and the camera 11B can simultaneously detect a three-dimensional object (an object such as the vehicle 60) on the road 70. The stereo camera 11 of the present embodiment is not limited to an imaging unit that captures a still image, and may be an imaging unit that captures a moving image.

  The vehicle type determination device 31 performs vehicle type determination based on the camera image acquired from the stereo camera 11, the vehicle weight of the vehicle 60 obtained by the vehicle weight meter 40, and the like.

(Functional configuration example of vehicle type identification device)
Next, a functional configuration example of the above-described vehicle type identification device 31 will be described. FIG. 10 is a diagram illustrating an example of a functional configuration of the vehicle type identification device according to the second embodiment.

  10 includes a storage unit 300, an acquisition unit 301, an image feature extraction unit 302, a distance measurement unit 303, a reference section change amount estimation unit 304, a state determination unit 305, and a vehicle entire view. A restoration unit 306, a number counting processing unit 307, an outline information acquisition unit 308, a crane detection unit 309, a tractor rear space measurement unit (space measurement unit) 310, a tow bar detection unit 311, a matching unit 312, The vehicle type determination unit 313, the network connection unit 314, and the control unit 315 are included.

  The storage unit 300 stores, for example, various programs for executing vehicle type determination processing, various setting information, and the like. Note that the information stored in the storage unit 300 may be acquired from other devices connected via a communication network represented by the Internet or a LAN, for example.

  In addition, the storage unit 300 stores third shape information that three-dimensionally represents the characteristic part of the vehicle component in the side surface direction of the vehicle in the three-dimensional coordinate system. For example, the storage unit 300 shows a crane part shape mounted on a crane vehicle, a partial image of a tractor part and a trailer part characterizing a tow vehicle, an index bar for towing a trailer part, etc. as a vehicle part. 3 shapes are stored. In the present embodiment, an example in which the characteristic part of the vehicle part stored in the storage unit 300 is stored as a three-dimensional surface model will be described. However, if the characteristic part of the vehicle part is represented three-dimensionally, Any model is acceptable.

  The storage unit 300 stores a crane part group 401 having a three-dimensional shape when the crane vehicle is viewed from the side as an example of the third shape information. FIG. 11 is a diagram illustrating a group of three-dimensional crane parts according to the second embodiment. As shown in FIG. 11, the three-dimensional crane parts group 401 stores one or a plurality of three-dimensional crane parts 402 that specify the type of crane vehicle. The crane parts group includes, but is not limited to, some characteristic information such as the shape and arrangement of the vehicle frame for characterizing the crane vehicle, as shown in FIG.

  In addition, the storage unit 300 stores a partial image database (DB) 501 viewed from the side of the vehicle as an example of the third shape information. FIG. 12 is a diagram illustrating a partial image DB viewed from the side of the vehicle according to the second embodiment. As shown in FIG. 12, the partial image DB 501 viewed from the side of the vehicle includes a partial image (three-dimensional shape) representing the side of the vehicle including a trailer portion 502 for distinguishing a general vehicle from a vehicle having a trailer. (The configuration pattern) is stored (hereinafter, the trailer portion 502 is referred to as “configuration pattern 502” as necessary). The partial image is an image including a partial region of the vehicle, such as a driver seat portion of the tractor portion that is the head portion of the vehicle, a cargo bed portion that loads the cargo of the container of the trailer portion that is the subsequent portion of the vehicle, and the like. Although there is, it is not limited to this.

  In addition, the storage unit 300 stores, as an example of the third shape information, a tow bar component group 601 having a three-dimensional shape when the tow vehicle is viewed from the side. FIG. 13 is a diagram illustrating a tow bar part group having a three-dimensional shape according to the second embodiment. As shown in FIG. 13, the three-dimensional tow bar part group 601 stores a three-dimensional index bar 602 that identifies the vehicle type of the vehicle that pulls one or more trailers. The tow bar part group is a part between the vehicle and the object to which the vehicle is towed, and looks like a bar in a captured image from the side. Whether the part is a tow bar is determined from the position and orientation of the tow bar and the relationship between the front and rear objects. In addition, the number of tow bars is also determined. Based on the determination result of the tow bar, it is possible to determine whether the target vehicle is a tow vehicle.

  The acquisition unit 301 acquires information obtained from each of the stereo camera 11 and the vehicle weight meter 40. For example, the acquisition unit 301 acquires an image obtained by capturing the side surface of the vehicle with the stereo camera 11. When the stereo camera 11 can capture a moving image, the acquisition unit 301 acquires an image (hereinafter referred to as “frame image”) obtained by capturing the side surface of the vehicle in units of frames.

  When the image feature extraction unit 302 acquires an image taken by the stereo camera 11 from the acquisition unit 301, the image feature extraction unit 302 extracts a three-dimensional object on the road 70. For example, the image feature extraction unit 302 extracts features such as the outer shape of the vehicle 60 from the image of the stereo camera 11.

  When the distance measurement unit 303 acquires images captured by the camera 11A and the camera 11B from the acquisition unit 301, the distance measurement unit 303 performs geometric calculation from the parallax information obtained by the matching processing of the images. Thereby, the distance measurement unit 303 measures, for example, a distance value from the position of the stereo camera 11 to each position on the side surface of the vehicle 60 (a position for each feature on the side surface of the vehicle 60 extracted from the image). .

  The reference interval change amount estimation unit 304 divides the imaging region represented in the frame image by a reference interval indicating a predetermined region. The reference section change amount estimation unit 304 measures the movement amount of the feature on the side surface of the vehicle within the reference section divided for each frame image. Thereby, the reference section change amount estimation unit 304 estimates the change amount of the vehicle between the frame images.

  That is, in the present embodiment, the reference section divided for each frame image by the reference section change amount estimation unit 304 is connected in consideration of the movement amount of the vehicle, so that a panoramic image showing the entire vehicle is obtained. Can be generated.

  The state determination unit 305 grasps the position of the vehicle 60 from the entry to the exit of the vehicle 60 as a state change according to the amount of change of the vehicle 60 in each reference section.

  The vehicle full view restoration unit 306 tracks the transition of each feature of the vehicle 60 extracted from each frame image captured when the vehicle is passing, according to the movement of the vehicle 60, and superimposes it on the traveling direction of the vehicle 60. Full-view information that three-dimensionally represents the side surface is generated.

  Specifically, the vehicle panoramic view restoration unit 306 collates the feature points of the immediately preceding and immediately following frame images while scanning typical feature amounts such as a vertical edge and a circle of the vehicle 60 in the traveling direction of the vehicle.

  Further, the vehicle whole view restoration unit 306 refers to the distance value obtained by the distance measurement unit 303 for each feature point of the frame image, and thereby indicates the position of each feature point on the side surface of the vehicle 60. The whole view information having the coordinates at (for example, distance coordinates from the origin with the vicinity of the stereo camera 11 as the origin) is reproduced. Thus, the full view information of the vehicle 60 indicates the three-dimensional shape on one side of the side surface of the vehicle 60 captured by the stereo camera 11.

  The number counting processing unit 307 determines that one vehicle 60 has passed when the state change of the presence position of the vehicle 60 grasped by the state determination unit 305 is determined to be “exit”. As a specific method, for example, the method described in Japanese Patent No. 5561414 can be used, but the method is not limited thereto.

  The outline information acquisition unit 308 is calculated from a set of feature points represented in a three-dimensional coordinate system, which is calculated from distance coordinates with each feature point of the vehicle 60 obtained from the stereo camera 11 based on the whole view information of the vehicle 60. For example, a plane projection transformation is used to generate a three-dimensional surface model (hereinafter referred to as “outline information”) that represents the outline of the vehicle 60 as a three-dimensional surface shape.

  That is, this embodiment makes it possible to identify the vehicle type with high accuracy by identifying the vehicle type with the outline information indicating the three-dimensional shape of the vehicle 60. For example, the matching of the three-dimensional shape is performed between the outline information indicating the three-dimensional shape of the vehicle 60 and the third shape indicating the three-dimensional shape of the vehicle component stored in the storage unit 300. As a result, it is possible to specify vehicle parts with high accuracy. This realizes highly accurate vehicle type identification.

  The crane detection unit 309 detects a crane portion mounted on a construction vehicle such as a crane vehicle. The crane detection unit 309 uses the outline information obtained by the outline information acquisition unit 308, for example, and a part of the vehicle is present at a predetermined height or more from the ground in a state where it does not contact the ground in a region in front of the vehicle body. In some cases, a region including a part thereof is detected as a crane part. Note that the crane detection unit 309 may detect the crane portion by matching the entire view information of the vehicle 60 generated by the vehicle entire view restoration unit 306 with the crane parts group 401, for example.

  The tractor rear space measurement unit 310 measures a space included in the entire vehicle view. The tractor rear space measurement unit 310 uses, for example, outline information obtained by the outline information acquisition unit 308, and corresponds to a space between the head portion of the vehicle 60 and the subsequent portion of the vehicle 60 (space at the rear of the tractor). Measure. Further, when the measured space is equal to or greater than a predetermined threshold, the tractor rear space measurement unit 310 uses the vehicle front portion as the tractor portion and the vehicle subsequent portion as the trailer portion or cargo portion of the vehicle outline information. As separate.

  The tow bar detection unit 311 detects the tow bar from the outline information obtained by the outline information acquisition unit 308, for example.

  The tow bar detection unit 311 has a contour (three-dimensional surface model) of the vehicle 60 that corresponds to a gap corresponding to, for example, a trailer among the three-dimensional surface models indicated by the contour information, and is equal to or greater than a predetermined threshold. Extract a changing region that changes. The tow bar detection unit 311 detects the tow bar from the change area. The predetermined threshold is determined based on the embodiment. The tow bar detection unit 311 may determine the measurement distance near the extracted change region using the measurement result of the distance measurement unit 303 and detect the tow bar based on the determination result.

  The matching unit 312 sets a region from the front end to the rear end of the trailer unit separated by the tractor rear space measurement unit 310 as a gaze region, and matches the configuration pattern registered in the partial image DB 501 viewed from the side of the vehicle. I do. The matching unit 312 calculates a matching score of the partial image. Even when there is a combination of partial images, it may be registered in advance.

  The vehicle type discriminating unit 313 discriminates the vehicle type based on the detection results obtained by the detection units described above and the matching results obtained by the matching unit 312. In addition, the vehicle type discrimination | determination part 313 may acquire shape information, such as the vehicle length of the discriminated vehicle type, using the outline information mentioned above.

  The network connection unit 314 transmits and receives data between the stereo camera 11 and the vehicle weight meter 40 via the communication network 50. The network connection unit 314 may be connected to a management server or the like via the Internet, a LAN, or the like. The network connection unit 314 transmits the vehicle type discrimination result acquired by the vehicle type discrimination device 31 to the management server, or the matching described above from the management server. It is also possible to acquire a pattern or the like used in.

  The control unit 315 controls the entire components of the vehicle type identification device 31. Further, the control unit 315 may perform control such as start and end of various processes in the second embodiment, control when an error occurs, and the like.

  In the second embodiment, the vehicle type determination unit 313 may determine whether or not the vehicle weight obtained from the acquisition unit 301 is equal to or less than the limit weight of the vehicle type determined by the vehicle type determination. In this case, when the vehicle weight acquired from the acquisition unit 301 exceeds the limit weight of the vehicle type, it is possible to notify the management server or the like as a vehicle that violates the regulations.

(Example of vehicle type discrimination process in the second embodiment)
Next, an example of the vehicle type determination process in the second embodiment will be described. FIG. 14 is a flowchart illustrating an example of a vehicle type determination process according to the second embodiment.

  As illustrated in FIG. 14, the image feature extraction unit 302 of the vehicle type identification device 31 extracts features such as the outer shape of a three-dimensional object (vehicle) from an image captured by the stereo camera 11 (S20). Next, the distance measuring unit 303 performs matching processing on the images obtained from the cameras 11A and 11B, performs geometric calculation from the parallax information, and, for example, calculates the distance from the position of the stereo camera 11 to each position on the side surface of the vehicle. Measure (S21).

  Next, the reference section change amount estimation unit 304 measures the movement amount of the feature on the side surface of the vehicle within the reference section divided for each frame image. Thereby, the variation | change_quantity of the vehicle between frame images is estimated (S22). Next, the state determination part 305 grasps | ascertains the presence position from the approach of the vehicle to the exit as a state change (S23).

  Next, the vehicle panoramic view restoration unit 306 generates the three-dimensional vehicle panoramic information by indicating the feature points on the side surface of the vehicle with the position coordinates of the three-dimensional coordinate system (S24). In the process of S24, the generated entire view information may be stored in the storage unit 300.

  Next, the number counting processing unit 307 determines that one vehicle has passed when it is determined that the state change of the vehicle presence position grasped by the state determination unit 305 is “exit”. A vehicle number counting process is performed (S25).

  The number counting processing unit 307 determines whether one vehicle has passed as a result of the processing of S25 (S26). When the number counting processing unit 307 determines that one vehicle has not passed (No in S26), the process returns to the process of S20.

  In addition, when the number counting processing unit 307 determines that one vehicle has passed (Yes in S26), the outline information acquisition unit 308 determines that the one vehicle based on the entire view information until one vehicle passes. The outline information representing the three-dimensional surface shape of the side outline of the vehicle is acquired (S27).

  Next, the crane detection unit 309 determines whether there is a crane part using the outline information acquired in the process of S27 (S28). When the crane detection unit 309 determines that there is no crane part (No in S28), the tractor rear space measurement unit 310 corresponds to the gap between the head part of the vehicle and the subsequent part of the vehicle using the outline information. The space is measured (S29).

  Next, the tractor rear space measurement unit 310 determines whether there is a space that is equal to or greater than a predetermined threshold among the spaces measured in the process of S29 (S30). When the tractor rear space measurement unit 310 determines that there is a space that is equal to or greater than the predetermined threshold (Yes in S30), the matching unit 312 performs matching on the gaze region separated as a trailer unit by the tractor rear space measurement unit 310. Perform (S31).

  When tractor rear space measurement unit 310 determines that there is no space equal to or greater than the predetermined threshold (No in S30), tow bar detection unit 311 uses the contour information to determine that the vehicle contour is equal to or greater than the predetermined threshold. The tow bar is detected by extracting the changing region that changes (S32).

  When it is determined that there is a crane part in the process of S28 described above (Yes in S28), after the process of S31 or after the process of S32, the vehicle type determination unit 313 determines the vehicle type based on each result. (S33), and the process ends.

(Specific example of the second embodiment)
(Detection of crane part)
Next, a specific example of the above-described second embodiment will be described. FIG. 15 is a diagram illustrating a specific example of crane part detection.

  The crane detection unit 309 detects, for example, a crane portion mounted on the front side of a construction vehicle 701 such as a crane vehicle as illustrated in FIG. As shown in FIG. 15B, the crane detection unit 309 detects a crane portion 702 that exists at a predetermined height or higher from the ground in a state where it does not contact the ground in a region 703 ahead of the vehicle body of the construction vehicle 701, for example. To detect.

  As a vehicle similar to a crane vehicle, a truck on which pole-shaped parts are mounted is conceivable. When comparing the pole-shaped part and the crane part, since the size of the crane part is large, for example, by referring to the information on the shape of the crane part measured by the distance measuring unit 303, the pole-shaped part and the crane part are It may be distinguished.

  Further, the crane detection unit 309 may detect the crane part by matching with the crane part group 402 shown in FIG. 11 using, for example, the vehicle whole view information generated by the vehicle whole view restoration unit 306.

  In addition, when discriminating a crane vehicle, you may match the vehicle front surface of a crane vehicle, for example using a monocular camera. For example, as shown in FIG. 11, the driver's seat portion of the crane vehicle is asymmetrical compared to other vehicles, such as a pattern 403 in which a cockpit is present on one side and a pattern 404 in which a seat is present at a lower position. Sex exists. Therefore, the difference with other vehicles is large.

  That is, matching accuracy is improved because there are few similar patterns. Therefore, it is also possible to determine that the vehicle is a crane vehicle by matching the tractor portion of the crane vehicle.

(Space measurement at the rear of the tractor and matching of the trailer)
FIG. 16 is a diagram illustrating an example of space measurement at the rear of the tractor. The tractor rear space measurement unit 310 is calculated from, for example, the entire view information of the image generated based on the feature points 806 such as the vertical edges and circles of the vehicle extracted from the frames 801 to 805 of the large vehicle shown in FIG. Use the outline information.

  Specifically, the tractor rear space measurement unit 310 includes, as shown in FIG. 16B, the space corresponding to the gap between the leading part and the subsequent part of the large vehicle (the tractor rear part). Space) 807 is measured. For example, when the measured space 807 is equal to or greater than a predetermined threshold, the tractor rear space measurement unit 310 separates the head portion as a tractor portion and the subsequent portion as a trailer portion or a cargo portion.

  The matching unit 312 sets a region from the front end to the rear end of the trailer unit separated by the tractor rear space measurement unit 310 as a gaze region, and is registered in the partial image DB 501 viewed from the side as shown in FIG. 12 described above. Matching with the configuration pattern 502 is performed.

  The matching unit 312 calculates the matching score of the partial image described above. As a result, the vehicle type determination unit 313 can determine, for example, a vehicle 503 and a general vehicle 504 having a trailer to be pulled as shown in FIG.

(Index bar detection)
FIG. 17 is a diagram illustrating an example of detection of a tow bar. For example, the tow bar detection unit 311 uses outline information calculated from full-view information of an image generated based on a vehicle feature point 1006 extracted from moving image frames 1001 to 1005 when the vehicle is shown in FIG. .

  The tow bar detection unit 311 tracks the coordinate sequence which is the outline information described above, and the change area where the outline (three-dimensional surface model) changes by a predetermined threshold or more in the gap 1007 between the trailers shown in FIG. The tow bar is detected by extracting. In addition, the tow bar detector 311 measures the number of tow bars present in the vehicle.

  The vehicle type discriminating unit 313 refers to the tow bar part group 601 shown in FIG. 13 described above, and discriminates the vehicle type as a doubles or a tow vehicle 603 if the number of tow bars is one. The vehicle type determination unit 313 determines the vehicle type as a full trailer 604 if the number of tow bars is two or more. Further, the vehicle type determination unit 313 determines the vehicle type as a normal large truck 605 when there is no gap or index bar in the tractor unit described above.

  The vehicle type identification device 31 according to the second embodiment described above generates vehicle panoramic information from an image acquired from the stereo camera 11, and based on the generated panoramic information, a large vehicle such as a crane portion, a trailer unit, an index rod, or the like. A characteristic part shape when viewed from the side is detected. As a result, it is possible to improve the accuracy of vehicle type discrimination for large vehicles.

  Further, when the above-described mechanism is combined with vehicle weight information, an overloaded vehicle can be found, and further, it can be extended to a general road in consideration of road pricing.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 1, 2 ... Vehicle type discrimination | determination system, 10 ... Monocular camera (1st imaging part), 11 ... Stereo camera (2nd imaging part), 20 ... Vehicle detector, 30, 31 ... Vehicle type discrimination | determination apparatus (information processing apparatus) , 40 ... vehicle weight scale, 50 ... communication network, 60 ... vehicle, 61 ... front region, 62 ... rear region, 70 ... road, 90, 300 ... storage unit, 91, 301 ... acquisition unit, 92 ... vehicle detection unit, 93 ... Tractor part detection part, 94 ... Trailer part cutout part (cutout part), 95, 312 ... Matching part, 96, 313 ... Vehicle type discrimination part, 97, 314 ... Network connection part, 98, 315 ... Control part, 302 ... Image feature extraction unit, 303 ... Distance measurement unit, 304 ... Reference section change amount estimation unit, 305 ... State determination unit, 306 ... Vehicle whole view restoration unit, 307 ... Number counting processing unit, 308 ... Outline information acquisition unit, 309 ... Crane inspection Department, 310 ... tractor rear space measuring section, 311 ... drawbar detector.

Claims (18)

A storage unit that stores first shape information representing the shape of a loading platform, luggage, or passenger car existing behind the vehicle;
An acquisition unit for acquiring an image of the vehicle imaged from the imaging unit;
A loading platform that exists behind the vehicle by performing collation using the rear region of the vehicle represented in the image obtained from the acquisition unit and the first shape information stored in the storage unit A vehicle type discriminating unit for discriminating the vehicle type of the vehicle from a luggage or a passenger car;
A vehicle type identification device. The storage unit stores second shape information representing the appearance of the front area of the vehicle based on the approach speed of the vehicle at the time of shooting by the imaging unit and the parameter information of the imaging unit,
Applying the second shape information to the front area of the vehicle represented in the image acquired by the acquisition section, and having a cutout section that cuts out the rear area of the vehicle other than the area,
The vehicle type determination unit performs collation using a rear region of the vehicle cut out by the cut-out unit and the first shape information stored in the storage unit.
The vehicle type identification device according to claim 1. The vehicle type discrimination unit
Performing the collation based on a combination of the shape of the loading platform, luggage, or passenger car represented by the first shape information stored in the storage unit,
The vehicle type identification device according to claim 1. The vehicle type of the vehicle includes information on one or more shapes of a width, a height, and a length of the vehicle.
The vehicle type identification device according to claim 1. A storage unit that stores third shape information that three-dimensionally represents the side surfaces of the components that constitute the vehicle;
An acquisition unit that acquires a plurality of images obtained by imaging a side surface of the vehicle by a plurality of imaging units;
Based on the plurality of images acquired by the acquisition unit, a distance measurement unit that measures the distance to each position on the side surface of the vehicle;
Based on the distance information obtained by the distance measurement unit and the plurality of images, a vehicle full view restoration unit that generates full view information of the vehicle that three-dimensionally represents a side surface of the vehicle;
Based on the entire view information of the vehicle generated by the vehicle entire view restoration unit and the third shape information stored in the storage unit, the vehicle type of the vehicle is determined from each component constituting the vehicle. A vehicle type discriminating unit,
A vehicle type identification device. An external information acquisition unit that acquires outline information indicating the outline of the side surface of the vehicle as a three-dimensional surface shape based on the entire view information of the vehicle;
The vehicle type identification device according to claim 5. Based on the outer information of the vehicle, the component constituting the vehicle has a crane detection unit that detects a crane unit having a predetermined height from the ground in a state where the vehicle is not in contact with the ground.
The vehicle type identification device according to claim 6. Based on the outer information of the vehicle, the component constituting the vehicle has a space measurement unit that measures a predetermined space or more between a head portion of the vehicle and a subsequent portion of the vehicle,
The vehicle type identification device according to claim 6. Tow bar detection for detecting a tow bar of the vehicle as the part constituting the vehicle by extracting a change region where the outer wall of the vehicle changes by a predetermined threshold or more based on the outer information of the vehicle Having a part,
The vehicle type identification device according to claim 6. An acquisition step of acquiring an image of the vehicle imaged from the imaging unit;
A rear region of the vehicle represented in the image obtained in the acquisition step, and first shape information representing a shape of a loading platform, a baggage, or a passenger vehicle existing in the rear of the vehicle stored in advance in a storage unit; The vehicle type determination step of determining the vehicle type of the vehicle from a loading platform, luggage, or passenger vehicle existing behind the vehicle,
A method for discriminating vehicle types. Based on the approach speed of the vehicle at the time of imaging by the imaging unit and the parameter information of the imaging unit, the second shape information representing the appearance of the front area of the vehicle stored in the storage unit is obtained in advance Cutting out a rear region of the vehicle other than the region by applying it to the front region of the vehicle represented in the acquired image,
The vehicle type determination step performs collation using the rear region of the vehicle cut out by the cutting step and the first shape information stored in the storage unit.
The vehicle type identification method according to claim 10. The vehicle type discrimination step includes
Performing the collation based on a combination of the shape of the loading platform, luggage, or passenger car represented by the first shape information stored in the storage unit,
The vehicle type identification method according to claim 10. The vehicle type of the vehicle includes information on one or more shapes of a width, a height, and a length of the vehicle.
The vehicle type identification method according to claim 10. An acquisition step of acquiring a plurality of images obtained by imaging a side surface of the vehicle by a plurality of imaging units;
Based on the plurality of images acquired in the acquisition step, a distance measurement step of measuring a distance to each position on the side surface of the vehicle;
Based on the distance information obtained in the distance measuring step and the plurality of images, a vehicle full view restoration step for generating full view information of the vehicle that three-dimensionally represents a side surface of the vehicle;
Based on the full view information of the vehicle generated in the full view restoration step of the vehicle, and the third shape information that three-dimensionally represents the side surfaces of the parts constituting the vehicle stored in advance in the storage unit, From each part constituting the vehicle, a vehicle type determination unit for determining the vehicle type of the vehicle,
A method for discriminating vehicle types. An external information acquisition step of acquiring outline information indicating the outline of the side surface of the vehicle as a three-dimensional surface shape based on the entire view information of the vehicle;
The vehicle type identification method according to claim 14. A crane detection step of detecting a crane portion having a height higher than a predetermined level from the ground in a state where the vehicle is not in contact with the ground, based on the outer information of the vehicle.
The vehicle type identification method according to claim 15. A space measurement step of measuring a space of a predetermined amount or more between a head portion of the vehicle and a subsequent portion of the vehicle as the component constituting the vehicle based on the outer information of the vehicle;
The vehicle type identification method according to claim 15. Tow bar detection for detecting a tow bar of the vehicle as the part constituting the vehicle by extracting a change region where the outer wall of the vehicle changes by a predetermined threshold or more based on the outer information of the vehicle Having steps,
The vehicle type identification method according to claim 15.

Images