JP2016062369A - Vehicle detection system, adjustment support device, vehicle detection method and adjustment method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a vehicle detection system, an adjustment support device, a vehicle detection method and an adjustment method capable of precisely detecting vehicles from images.SOLUTION: The vehicle detection system includes: an imaging unit; a geometrical conversion part; and a vehicle detection part. The imaging unit acquires images of a road by taking images of the road. The geometrical conversion part performs a geometry conversion based on the installation position and the direction of the imaging unit with respect to plural focusing areas in the images in which images of vehicles travelling on the road are taken. The vehicle detection part detects the vehicles from the images in the focusing areas after the geometry conversion by the geometrical conversion part.SELECTED DRAWING: Figure 10

Description

  Embodiments described herein relate generally to a vehicle detection device, an adjustment support device, a vehicle detection method, and an adjustment method.

Conventionally, an image of a road is picked up using an image pickup device such as a camera, and a vehicle traveling on the picked up image is detected from an acquired image. By detecting this vehicle, counting of passing vehicles and acquisition of number information of passing vehicles are performed. One method for detecting a vehicle from such an image is pattern matching. It is possible to detect the vehicle with high accuracy by matching the acquired image with the pattern image.
However, when vehicle detection is performed by pattern matching, the detection accuracy greatly depends on the symmetry and distortion of the detection target captured in the image. Therefore, depending on the installation position and the installation angle of the imaging device, the matching accuracy may be greatly reduced.

JP2013-117768A JP 2010-218438 A JP 2012-3708 A

  The problem to be solved by the present invention is to provide a vehicle detection device, an adjustment support device, a vehicle detection method, and an adjustment method that can detect a vehicle from an image with higher accuracy.

  The vehicle detection device according to the embodiment includes an imaging unit, a geometric conversion unit, and a vehicle detection unit. The imaging unit captures a road and acquires an image of the road. The geometric conversion unit performs geometric conversion according to the installation position and orientation of the imaging unit for a plurality of gaze regions in the image where the vehicle traveling on the road is captured. A vehicle detection part detects the said vehicle from the image of the gaze area | region geometrically transformed by the said geometry conversion part.

The system configuration figure showing the system configuration of vehicle detection system 1 of an embodiment. The figure which shows the example of installation of the vehicle detection apparatus 3. FIG. The figure which shows the example of installation of the vehicle detection apparatus 3. FIG. The figure which shows an example of the method of judging passage of a vehicle from an image. The figure which shows an example of the method of judging passage of a vehicle from an image. The figure which shows an example of the method of judging passage of a vehicle from an image. The figure which shows the outline | summary of the method by which the vehicle detection apparatus 3 of embodiment determines the passage of a vehicle. The figure which shows the outline | summary of the method by which the vehicle detection apparatus 3 of embodiment determines the passage of a vehicle. The figure which shows the outline | summary of the method by which the vehicle detection apparatus 3 of embodiment determines the passage of a vehicle. FIG. 3 is a functional block diagram showing a functional configuration of the vehicle detection device 3. The figure which shows the specific example of geometric transformation. FIG. 3 is a functional block diagram showing a functional configuration of the adjustment support apparatus 2. The figure which shows the example of the division | segmentation of a road area. The figure which shows the specific example of a gaze area | region. The figure which shows the specific example of a gaze area | region. The figure which shows the specific example of a gaze area | region. The figure which shows the specific example of the gaze area | region in the case of imaging a vehicle from the back. The figure which shows the specific example of the gaze area | region in the case of imaging a vehicle from the back. The flowchart which shows the flow of the vehicle detection process by the vehicle detection apparatus 3 of embodiment. The figure which shows the specific example of the priority table 321. FIG. The figure which shows the specific example of the screen for confirming the recognition result of a license plate. The figure which shows the specific example of the position information table 322.

Hereinafter, a vehicle detection device, an adjustment support device, a vehicle detection method, and an adjustment method according to embodiments will be described with reference to the drawings.
FIG. 1 is a system configuration diagram illustrating a system configuration of a vehicle detection system 1 according to the embodiment.
The vehicle detection system 1 includes an adjustment support device 2 and a vehicle detection device 3.
The adjustment support device 2 is a device that supports installation or adjustment work (hereinafter referred to as “adjustment work”) of the vehicle detection device 3. For example, the adjustment support device 2 is configured using a terminal device such as a PC or a smartphone. An application (hereinafter referred to as “adjustment tool”) for adjusting the vehicle detection device 3 is installed in the adjustment support device 2. The operator can confirm information such as setting information and adjustment results by operating the adjustment tool. The adjustment support device 2 includes a communication interface such as a LAN (Local Area Network) and is connected to the vehicle detection device 3 when the vehicle detection device 3 is installed or adjusted. The adjustment support device 2 may be directly connected to the vehicle detection device 3 or may be connected via a network. Further, the connection of the adjustment support device 2 to the vehicle detection device 3 may be a wired connection or a wireless connection.

  The vehicle detection device 3 is a device that detects a vehicle traveling on a road. The vehicle detection device 3 is installed at a predetermined position above the road and images the road. The vehicle detection device 3 detects a vehicle from image data acquired by imaging. The vehicle detection device 3 determines the traffic situation of the vehicle based on the detection result, and counts the vehicles that have passed. The vehicle detection device 3 reads the number from the detected vehicle image. The vehicle detection device 3 transmits information such as the presence / absence of vehicle detection, the traffic status of the vehicle, the number of vehicles, and the number acquired based on the detection to other devices and systems. These pieces of information are collectively referred to as detection information. In the case of the example in FIG. 1, the vehicle detection device 3 notifies the detection information to the host system 4.

  In addition, the information which the vehicle detection apparatus 3 transmits to the high-order system 4 is not limited to detection information. The vehicle detection device 3 may transmit any information held by the own device to the host system 4. For example, the vehicle detection device 3 may transmit the setting information of the own device, information acquired or generated in the course of the vehicle detection process, and the like to the upper system 4. Further, the vehicle detection device 3 may transmit various types of information acquired from the adjustment support device 2 to the host system 4. The vehicle detection device 3 may acquire information from the host system 4.

  The host system 4 is a system that receives detection information from the vehicle detection system 1. For example, the host system 4 may be a system that monitors traffic conditions or an ETC (Electronic Toll Collection) system that automates toll collection.

2 and 3 are diagrams illustrating an installation example of the vehicle detection device 3.
Reference numeral 5 in FIGS. 2 and 3 represents a vehicle traveling on the road. Reference numeral 6 denotes a structure provided on the side of the road in order to install the vehicle detection device 3 over the road. The vehicle detection device 3 is mounted on the structure 6 to image the road from the sky. FIG. 2 is an example in which the vehicle detection device 3 is installed so as to image a road in a direction opposite to the traveling direction of the vehicle 5. In this case, the vehicle detection device 3 images the entire surface of the vehicle. FIG. 3 is an example in which the vehicle detection device 3 is installed so as to image a road in the forward direction with respect to the traveling direction of the vehicle 5. In this case, the vehicle detection device 3 images the back surface of the vehicle.

4, 5 and 6 are diagrams illustrating an example of a method for determining the passage of a vehicle from an image.
4, 5, and 6 are methods disclosed in Japanese Patent Application Laid-Open No. H10-260260, and therefore only an overview is given here. In FIG. 4, images 101 to 104 are images in which vehicles passing through a road are captured in time series in the order of images 101 to 104. Reference numerals 105 and 106 are boundary lines that divide the images 101 to 104 into three in the direction of the traveling direction of the vehicle. The images 101 to 104 are divided into Area1, Area2, and Area3 by the boundary lines 105 and 106. In the image 101, vehicles are detected in Area1 and Area2. In the image 102 and the image 103, the vehicle is detected in Area1, Area2, and Area3. In the image 104, a vehicle is detected in Area3.

  FIG. 5 is a diagram illustrating transition of the states of Area1, Area2, and Area3 when the vehicle passes. The images 111 to 115 are divided into Area1, Area2, and Area3 from the left of the image, respectively. The traveling direction of the vehicle is the right direction of the page. First, there is no vehicle to be detected at the timing when the image 111 is captured. Then, at the timing when the image 112 is captured, a new vehicle is detected in Area1. The detected vehicle travels in the traveling direction, and is detected by Area1, Area2, and Area3 at the timing when the image 113 is captured. The detected vehicle further travels and is detected by Area 3 at the timing when the image 114 is captured. Then, the vehicle completely passes through Area3, and at the timing when the image 115 is captured, the vehicle to be detected does not exist again. The vehicle detection states shown in the images 111 to 115 are represented as S0, S1, S3, S2, and S0, respectively.

  FIG. 6 is a diagram showing vehicle detection state transitions in each reference section as ON or OFF when Area 1 is the reference section a and Area 2 is the reference section b. ON is a state indicating that there is a detected vehicle, and OFF is a state indicating that there is no detected vehicle. When the vehicle passes, the reference section a (Area 1) is turned on in the states S1 and S3 from FIG. Similarly, the reference section b (Area 2) is turned on in the states of S3 and S2. In other words, the passage of the vehicle can be determined by detecting that the state represented by ON or OFF of the reference section a and the reference section b has changed in the order of S0, S1, S3, S2, and S0.

  In the above method, since the state of S0 (no vehicle) is detected to determine the passage of the vehicle, in order to obtain high detection accuracy, it is necessary to image the vehicle from directly above or next to it. However, when a vehicle is imaged from directly above or from the side, it is difficult to image a license plate, and it is not possible to acquire number information for identifying the vehicle. Therefore, the vehicle detection device 3 according to the present embodiment images the vehicle from the front diagonally upper side or the rear diagonal upper side of the vehicle as shown in FIGS. 2 and 3.

7, 8, and 9 are diagrams illustrating an outline of a method in which the vehicle detection device 3 according to the embodiment determines the passage of the vehicle.
An image 200 in FIG. 7 is an image of a road imaged by the vehicle detection device 3. In the image 200, a region indicated by reference numeral 201 is a road. Hereinafter, an area indicating a road in the image is referred to as a road area. The road indicated by the road region refers to one or a plurality of lanes on which the vehicle travels, not a road in a wide range. The image 200 is an example of a one-lane road area. The areas indicated by reference numerals 202 and 203 are areas other than roads (hereinafter referred to as “non-road areas”). The arrow shown in the road area 201 indicates the traveling direction of the vehicle.

  FIG. 8 is a diagram illustrating a specific example of a gaze area set in the vehicle detection device 3. The gaze area is an area of an image used by the vehicle detection device 3 for vehicle detection. An image 200 in FIG. 8 is the same image as the image 200 in FIG. In the example of FIG. 8, gaze areas indicated by reference signs V1 to V5 are set in the order of the traveling direction of the vehicle. Each gaze area is formed of a plane that is orthogonal to the traveling direction of the vehicle and the road surface and has a width that is the same as the width of the road. In the case of the example in FIG. 8, the vehicle detection device 3 images the road from the upper front side of the vehicle, so the shape of the gaze area is not a rectangle but a trapezoid as shown in the figure.

  An image 210 in FIG. 9 is an image in which a state where the vehicle travels on the road imaged in the image 200 is captured. The vehicle detection device 3 performs image processing on the images of the gaze areas V1 to V5, and detects the front portion of the vehicle including the license plate from each gaze area. In the case of FIG. 9, the front portion of the vehicle is detected in the gaze region V4. The vehicle detection device 3 determines the passage of the vehicle based on the transition of the gaze area where the front portion of the vehicle is detected. Specifically, the vehicle detection device 3 first detects the front portion of the vehicle in the gaze region V1. When the vehicle detection device 3 detects the vehicle in the gaze area V1, the vehicle detection device 3 starts tracking the detected vehicle. After detecting the vehicle in the gaze area V1, the vehicle detection device 3 determines that the vehicle has passed when detecting the vehicle in the order of the gaze areas V2, V3, V4, and V5.

  In addition, when detecting the vehicle from the image of the gaze area, the vehicle detection device 3 may acquire the detected feature of the vehicle by a technique such as Hough conversion. The vehicle detection device 3 can more accurately determine the passage of the vehicle by determining whether or not the vehicle detected in a certain gaze area is the vehicle detected last time in the gaze area. The vehicle detection device 3 may acquire information such as the width, height, and position of the license plate as vehicle characteristics. The vehicle detection device 3 reads the number plate from the image of the vehicle detected in any of the gaze areas from the start of tracking the vehicle until the vehicle passes, and acquires the vehicle number information.

Note that the size of the gaze area on the image increases as the gaze area is located closer to the vehicle detection device 3. Therefore, in the example of FIG. 9, the gaze area becomes larger in the order of the gaze areas V1, V2, V3, V4, and V5. The larger the size of the gaze area, the more the gaze area includes more pixels. Therefore, the subject is imaged more clearly. Conversely, the smaller the size of the gaze area, the fewer pixels included in the gaze area. Therefore, the sharpness of the subject is reduced. On the other hand, in the gaze area at a position close to the vehicle detection device 3, the vehicle is imaged from a direction more directly above or just beside. Therefore, in the gaze area near the vehicle detection device 3, the vehicle is clearly imaged, but it is difficult to read the license plate. For such a reason, the reading of the license plate is preferably performed in a gaze area at an intermediate position that is not too far from the vehicle detection device 3 and not too close. For example, in the case of the example of FIG. 9, the vehicle detection device 3 may read the license plate with the image of the gaze region V3 or V4.
Hereinafter, the details of the vehicle detection device 3 that performs vehicle detection by the above method will be described.

FIG. 10 is a functional block diagram showing a functional configuration of the vehicle detection device 3.
The vehicle detection device 3 includes a CPU (Central Processing Unit), a memory, an auxiliary storage device, and the like connected by a bus, and executes a vehicle detection program. The vehicle detection device 3 includes a communication unit 31, a storage unit 32, an imaging unit 33, a vehicle candidate detection unit 34, a gaze area correction unit 35, a vehicle detection unit 36, a vehicle state determination unit 37, and a vehicle counting unit by executing a vehicle detection program. 38 and a setting registration unit 39. All or some of the functions of the vehicle detection device 3 may be realized using hardware such as an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a field programmable gate array (FPGA). . The vehicle detection program may be recorded on a computer-readable recording medium. The computer-readable recording medium is, for example, a portable medium such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, or a storage device such as a hard disk built in the computer system. The vehicle detection program may be transmitted via a telecommunication line.

The communication unit 31 is configured using a communication interface such as a LAN. The communication unit 31 communicates with the adjustment support device 2.
The storage unit 32 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 32 stores in advance a pattern image for matching (hereinafter referred to as “dictionary data”) used in the vehicle detection process.
The imaging unit 33 is configured using an imaging device such as a camera. The imaging unit 33 images a road at a predetermined position and orientation, and generates image data. The imaging unit 33 outputs the image data to the vehicle candidate detection unit 34.

The vehicle candidate detection unit 34 acquires image data from the vehicle candidate detection unit 34 and the imaging unit 33. The vehicle candidate detection unit 34 performs image processing on the image data acquired by the vehicle candidate detection unit 34 and extracts feature points of the object captured in the image. The vehicle candidate detection unit 34 calculates the speed of the object indicated by the feature point from the same feature point in the continuous image data. The vehicle candidate detection unit 34 detects an object that is a vehicle candidate based on the calculated speed.
The gaze area correction unit 35 (geometric conversion unit) performs geometric conversion on the gaze area of the image in which the vehicle candidate is detected. This geometric conversion is a process of converting a length ratio on the image of the subject into a length ratio in real space.

FIG. 11 is a diagram illustrating a specific example of geometric transformation.
FIG. 11 shows geometric transformation of an image captured in the gaze region V4 of FIG. The image of the gaze region V4 ′ is an image obtained by geometrically converting the image of the gaze region V4. In this way, the gaze area correction unit 35 corrects the size of the subject on the image caused by the distance in the distance by geometrically converting the trapezoidal gaze area V4 into the rectangular gaze area V4 ′. A parameter (hereinafter referred to as “geometric parameter”) used for the geometric transformation is an angle indicating how much the gaze region V4 is inclined with respect to the gaze region V4 ′, and this angle is the angle of the gaze region V4. It can be acquired from the setting information.

Returning to the description of FIG.
The vehicle detection unit 36 detects the front portion of the vehicle by performing a matching process with dictionary data on the image of the gaze area corrected by the gaze area correction unit 35. In the dictionary data, a plurality of image data are prepared in advance according to characteristics such as the vehicle type, the size of the vehicle body, and the vehicle width. The vehicle detection unit 36 outputs the detection result to the vehicle state determination unit 37. Moreover, the vehicle detection part 36 detects a number plate from the image of the gaze area | region where the vehicle was detected, and reads a number.

The vehicle state determination unit 37 determines the progress of the vehicle based on the detection result of the vehicle detection unit 36. When it is determined that the vehicle has passed, the vehicle state determination unit 37 notifies the vehicle counting unit 38 and the host system 4 that the vehicle has passed.
The vehicle counting unit 38 counts passing vehicles based on the vehicle passing notification.
The setting registration unit 39 stores the setting of the own device transmitted from the adjustment support device 2 in the storage unit 32.

FIG. 12 is a functional block diagram showing a functional configuration of the adjustment support apparatus 2.
The adjustment support apparatus 2 includes a CPU, a memory, an auxiliary storage device, and the like connected by a bus, and executes an adjustment support program. The adjustment support device 2 functions as a device including the communication unit 21, the display unit 22, the input unit 23, the storage unit 24, the information acquisition unit 25, the adjustment information acquisition unit 26, and the setting unit 27 by executing the adjustment support program. In addition, the adjustment support apparatus 2 enables the operator to use the adjustment tool by executing the adjustment support program. Note that all or part of each function of the adjustment support apparatus 2 may be realized using hardware such as an ASIC, a PLD, or an FPGA. The adjustment support program may be recorded on a computer-readable recording medium. The computer-readable recording medium is, for example, a portable medium such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, or a storage device such as a hard disk built in the computer system. The adjustment support program may be transmitted via a telecommunication line.

The communication unit 21 is configured using a communication interface such as a LAN. The communication unit 21 communicates with the vehicle detection device 3.
The display unit 22 is configured using a display device such as a CRT (Cathode Ray Tube) display, a liquid crystal display, or an organic EL (Electro-Luminescence) display. Or the display part 22 may be comprised as an interface which connects these display apparatuses to an own apparatus. The display unit 22 displays information related to the adjustment work (hereinafter referred to as “adjustment information”) in accordance with the operation of the worker.

  The input unit 23 is configured using an input device such as a mouse, a keyboard, or a touch panel. Alternatively, the input unit 23 may be configured as an interface for connecting these input devices to the own device. The input unit 23 accepts an operator's operation and outputs input information to each functional unit according to the content of the operation.

  The storage unit 24 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 24 stores information related to the setting of the vehicle detection device 3 and information acquired from the vehicle detection device 3.

  The information acquisition unit 25 acquires information necessary for the adjustment work from the vehicle detection device 3. For example, the information necessary for the adjustment work includes information regarding the operation of the vehicle detection device 3 such as the detection result and detection position of the vehicle, acquired image data, and the like. The information acquisition unit 25 stores information acquired from the vehicle detection device 3 in the storage unit 24.

  The adjustment information acquisition unit 26 acquires various types of information from the storage unit 24 and generates adjustment information to be displayed on the display unit 22. The adjustment information acquisition unit 26 generates adjustment information according to the input information acquired from the input unit 23. The adjustment information acquisition unit 26 outputs the generated adjustment information to the display unit 22 for display.

  The setting unit 27 performs various settings related to the operation of the vehicle detection device 3. Specifically, the setting unit 27 acquires input information from the input unit 23. The setting unit 27 acquires setting information for various settings from the input information. The setting information includes information indicating the type of setting and a setting value. The setting unit 27 stores the acquired setting information in the storage unit 24. The setting unit 27 transmits the acquired setting information to the vehicle detection device 3 to reflect the setting in the vehicle detection device 3. Types of setting include installation setting, road area setting, image feature setting, and gaze area setting. The setting unit 27 includes an installation information setting unit 271, a road region setting unit 272, an image feature setting unit 273, and a gaze region setting unit 274 as functional units that perform various types of settings.

  The installation information setting unit 271 performs installation setting of the image detection device 3. The installation setting is a setting related to the installation position and installation state of the image detection device 3. Specifically, the installation information setting unit 271 sets the installation height, the depression angle, the viewing angle, and the like of the image detection device 3. Further, the installation information setting unit 271 may be configured to set position information such as latitude and longitude.

  The road area setting unit 272 sets the road area of the image detection device 3. The road area setting is a setting for registering a position in an image of an area where a road is imaged (hereinafter referred to as “road area”) in an image captured by the image detection device 3. The road here means a lane in which the vehicle travels. The image is classified into a road area and an area other than the road by the road area setting. The road area set by the road area setting unit 272 is used for gaze area setting. The position of the road region in the image may be detected by image processing, or may be designated by an operator input. In the present embodiment, the road area setting unit 272 acquires the position of the road area based on the input information of the worker. For example, the operator operates the input unit 23 to display the image captured by the vehicle detection device 3 on the display unit 22. The worker inputs the boundary line of the road area in the displayed image. The road area setting unit 272 approximates the road area represented by the input boundary line with a polygon.

  For example, when the image captured by the image detection device 3 is the image 200 in FIG. 7, the worker inputs a boundary line between the road region 201 and the non-road regions 202 and 203. In the example of the image 200, the road is close to a straight line. In this case, the road area setting unit 272 approximates the road area 201 with a pentagon by approximating the input boundary line with a straight line.

Returning to the description of FIG.
When the road area is identified, the road area setting unit 272 divides the road area into a plurality of areas. Specifically, the road area setting unit 272 divides the road area with a straight line orthogonal to the traveling direction of the vehicle.

FIG. 13 is a diagram illustrating an example of dividing a road area.
FIG. 13 shows an example in which the road area 201 in FIG. 7 is divided into six areas. In this case, for example, the worker inputs a sample of points dividing the road area 201 on the boundary line between the road area 201 and the non-road areas 202 and 203. In FIG. 13, examples of sample points on the right side of the road are represented by R1, R2,. Similarly, examples of sample points on the left side of the road are represented by L1, L2,. A set of sample points on the right side of the road is described as {R}. A set of sample points on the left side of the road is denoted as {L}. The road area setting unit 272 selects sample points from the sets {R} and {L} so that a straight line connecting the two points is orthogonal to the traveling direction of the vehicle. A set of sample points selected in this way is described as (Rn, Ln) (n is an integer of 1 or more). Further, in each set of sample points, a straight line having the sample points as end points is described as Rn-Ln. In the example of FIG. 13, it is shown that R1-L1, R2-L2, R3-L3, R4-L4, and R5-L5 were obtained as straight lines dividing the road region 201.

Returning to the description of FIG.
The image feature setting unit 273 sets the feature of an object existing in the imaging field of view of the vehicle detection device 3. For example, the image feature setting unit 273 sets information related to the position and shape of an object that is regularly imaged, such as a pillar or a streetlight. It is possible to detect a change in the installation state of the vehicle detection device 3 by comparing the characteristics set in this way between images captured in time series.

  The gaze area setting unit 274 sets an area (hereinafter referred to as “gaze area”) used for vehicle detection for the image acquired by the vehicle detection device 3. Specifically, the gaze area setting unit 274 sets a plane that includes the straight line Rn-Ln that divides the road area and is orthogonal to the road surface as the gaze area. For example, gaze areas set based on the straight lines R1-L1, R2-L2, R3-L3, R4-L4, and R5-L5 obtained in FIG. 13 are gaze areas V1 to V5 in FIG.

  As shown in FIG. 8, the height from the road surface is necessary for setting the gaze area. This height is determined by taking an image of a pole or the like whose length is known and comparing it with the distance on the image with this length as a reference. Here, the height of the real space region indicated by the gaze region is the same at any point on the straight line Rn-Ln. The vehicle detection device 3 detects the vehicle by performing a matching process with the pattern image held in advance for the image of the gaze area set in this way.

14, FIG. 15 and FIG. 16 are diagrams showing specific examples of the gaze area.
FIG. 14 is a diagram illustrating an example in which the distance between the gaze regions is set at an interval longer than the vehicle body length of the vehicle. An image 300 in FIG. 14 is an image of a situation in which two vehicles are traveling close to each other on the road imaged in the image 200. In the example of FIG. 14, the front portion of the preceding vehicle is detected in the gaze region V7. Further, for the following vehicle, the front portion of the vehicle is detected in the gaze region V6. Thus, by providing a plurality of gaze regions, the vehicle detection device 3 can detect a plurality of vehicles at the same time. However, if an excessively large gaze area is set, problems such as vehicle misdetection and performance degradation may occur. Therefore, the distance between the gaze regions may be set at an interval that is about the same as or longer than the vehicle body length of the vehicle.

  When the imaging target road has a plurality of lanes, different gaze areas may be set for each lane as shown in FIG. As shown in FIG. 15, by setting a gaze area for each lane, the vehicle detection device 3 can detect vehicles in a plurality of lanes with one image. However, when a gaze area is set for a plurality of lanes, the same vehicle may be captured in a plurality of gaze areas. For example, when a vehicle with a large vehicle body travels in the front lane, the vehicle may be imaged in the gaze area of the back lane. And when it is set so that a gaze area may overlap in an adjacent lane, this possibility increases more. For example, in the case of the example in FIG. 15, a vehicle imaged in the gaze area 401 may be imaged in the gaze area 402. That is, there is a possibility that the same vehicle is detected in a plurality of gaze regions, and such a region can be a factor that reduces the accuracy of the matching process. Therefore, the gaze area setting unit 274 may mask an area in the gaze area where the same vehicle is imaged twice and exclude the area from the area used for the matching process.

  In addition, depending on the type of vehicle, a portion that is a characteristic of the vehicle may not fit in the gaze area. Therefore, a plurality of height gaze areas may be set in the vehicle detection device 3. For example, in a vehicle with a high vehicle height as shown in FIG. 16, a part of the front portion of the vehicle may not fit in the gaze area, and the accuracy of the matching process may be reduced. Further, in a situation where the amount of light is small, such as at night, the edge information of the object in the image is greatly lost, and the accuracy of the matching processing may be reduced. In order to suppress such a decrease in accuracy of the matching process, a wider gaze area may be set. For example, in the case of FIG. 16, the gaze area 501 represents a normal gaze area. In this case, the gaze area may be set as a gaze area 502 as a wider gaze area. The gaze area 502 is obtained by enlarging the gaze area 501 in the height direction of the vehicle. The gaze area 503 is a difference area between the gaze area 501 and the gaze area 502. The matching process may be performed on the gaze area 502 or may be performed on each of the gaze areas 501 and 503.

  The gaze area has been described above by taking as an example the case where the vehicle is imaged from the front as shown in FIG. The gaze area described above may be applied when the vehicle is imaged from either the front or the back. On the other hand, the setting of the gaze area that is particularly effective when the vehicle is imaged from the back side will be described below.

17 and 18 are diagrams illustrating specific examples of the gaze area when the vehicle is imaged from the back.
Reference numerals 601 and 602 in FIG. 17 indicate the imaging range of the vehicle detection device 3. In FIG. 17, two vehicles indicated by reference numeral 603 represent the same vehicle. The vehicle 603 at position A travels to the left in the drawing and reaches position B. When the vehicle is imaged from the back as shown in FIG. 17, the vehicle detection device 3 cannot image the license plate on the front of the vehicle. Therefore, in this case, the vehicle detection device 3 detects the number plate on the rear surface of the vehicle.

  Images 701 and 702 in FIG. 18 are images obtained by capturing the progress of the vehicle from the back of the vehicle. In this case, the relationship between the direction of the vehicle detection device 3 and the traveling direction of the vehicle is as shown in FIG. The image 702 is an image captured at a time later than the time when the image 701 was captured. Therefore, in the image 702, the vehicle imaged in the image 701 is imaged at a position moved in the traveling direction (upper right in the drawing). Reference numerals 703 to 705 are gaze areas set in the vehicle detection device 3 that captures the images 701 and 702. The gaze regions 703 to 705 indicate regions at the same position in the images 701 and 702. In the image 701, the license plate on the back of the vehicle is imaged at the position of the gaze area 704, and in the image 702, the image is captured at the position of the gaze area 705.

  As described above, when the vehicle is imaged from the back side, the license plate may not appear in the gaze region 703 depending on the type and height of the vehicle, the position and orientation of the vehicle detection device 3. For example, for a vehicle such as a truck, there is a possibility that only the cargo bed portion of the vehicle is shown in the gaze area 703. In this case, the vehicle detection device 3 performs a matching process on a gaze area that has no possibility of detecting the license plate of the vehicle or has low detection accuracy even if it is detected. System resources are wasted. Therefore, the adjustment support apparatus 2 may perform a setting for masking such a gaze area as a target for the matching process in the vehicle detection apparatus 3.

FIG. 19 is a flowchart illustrating a flow of a vehicle detection process performed by the vehicle detection device 3 according to the embodiment.
First, the imaging unit 33 images a road and acquires image data (step S101). The imaging unit 33 outputs the acquired image data to the vehicle candidate detection unit 34. The vehicle candidate detection unit 34 acquires image data from the imaging unit 33. The vehicle candidate detection unit 34 performs preprocessing on the acquired image data (step S102). In the preprocessing, the vehicle candidate detection unit 34 first performs filter processing using a Sobel filter or the like to remove noise from the image data. This is because imaging in situations where the amount of light is low, such as at night, is susceptible to noise. And the vehicle candidate detection part 34 extracts the feature point of a moving body from the image from which noise was removed. Any existing method may be used as a method for extracting feature points of a moving object from an image.

The vehicle candidate detection unit 34 calculates a movement vector from the same feature point in continuous image data. The vehicle candidate detection unit 34 calculates a movement vector for each of the plurality of feature points, and calculates a movement vector group. The vehicle candidate detection unit 34 calculates a statistical value such as an average or a variance for the absolute value of each movement vector (step S103). The vehicle candidate detection unit 34 determines the presence or absence of a vehicle candidate by comparing the statistical value with a predetermined threshold (step S104). The vehicle candidate detection unit 34 determines that there is a vehicle candidate when the statistical value is equal to or greater than a predetermined threshold. When it is determined that there is a vehicle candidate (step S104—YES), the image data is output to the gaze area correction unit 35 together with the position where the vehicle candidate is detected. On the other hand, the vehicle detection unit 36 determines that there is no vehicle candidate when the statistical value is less than a predetermined threshold. When it is determined that there is no vehicle candidate (step S104-NO), the process returns to step S101, and the vehicle detection device 3 detects the vehicle candidate for the next image data.

  The gaze area correction unit 35 performs geometric conversion on the gaze area of the image determined to have a vehicle candidate (step S105). The gaze area correction unit 35 determines the gaze area to be converted based on the position of the image where the vehicle candidate is detected. The gaze area correction unit 35 converts the image of the gaze area using the geometric parameter calculated by the adjustment support apparatus 2. The gaze area correction unit 35 outputs an image of the gaze area corrected by the geometric transformation to the vehicle detection unit 36.

  The vehicle detection unit 36 performs matching processing with dictionary data on the corrected image of the gaze area (step S106). By this matching process, the vehicle is detected from the image of the gaze area. Specifically, the vehicle detection unit 36 calculates a score indicating the correlation between the gaze area image and the dictionary data image. At this time, the vehicle detection unit 36 performs matching processing while changing the resolution and size of the image of the dictionary data in order to improve matching accuracy. The vehicle detection unit 36 calculates a score for each of a plurality of images included in the dictionary data and for each change in resolution and size. The vehicle detection unit 36 determines the presence or absence of a vehicle by comparing the calculated score with a predetermined threshold (step S107). When the score is equal to or greater than a predetermined threshold (step S107—YES), the vehicle detection unit 36 determines that there is a vehicle and notifies the vehicle state determination unit 37 of the gaze area where the vehicle is detected. On the other hand, when the score is less than the predetermined threshold (step S107—NO), the vehicle detection unit 36 determines that there is no vehicle. When it is determined that there is no vehicle, the process returns to step S101, and the vehicle detection device 3 detects a vehicle candidate for the next image data.

  The vehicle state determination unit 37 receives a notice of the gaze area where the vehicle is detected from the vehicle detection unit 36. The vehicle state determination unit 37 determines the progress of the vehicle based on the notified gaze area and the previously notified gaze area (step S108). For example, the vehicle state determination unit 37 determines the vehicle progress status into three states of “entry”, “passing”, and “exit”. When the vehicle state determination unit 37 determines “exit”, the vehicle state determination unit 37 notifies the vehicle counting unit 38 that the vehicle has passed. The vehicle detection unit 36 detects the number plate from the image of the gaze area where the vehicle is detected, and reads the number (step S109). When the vehicle count unit 38 receives a notification that the vehicle has passed from the vehicle state determination unit 37, the vehicle count unit 38 adds 1 to the current number of vehicles (step S110). When the number of vehicles is added, the vehicle counting unit 38 notifies the host system 4 that a new vehicle has passed.

  In the vehicle detection system 1 of the embodiment configured as described above, the vehicle detection device 3 includes a gaze area correction unit 35 that corrects the gaze area in the vehicle detection process. By providing the gaze area correction unit 35, the vehicle detection device 3 can perform the matching process between the gaze area image and the dictionary data image with higher accuracy. As a result, the vehicle detection accuracy of the vehicle detection device 3 is improved.

  In the vehicle detection system 1 according to the embodiment, an operator who installs or adjusts the vehicle detection device 3 can visually check information regarding the installation by working using the adjustment support device 2. Therefore, the accuracy of installation or adjustment of the vehicle detection device 3 is improved, and the vehicle detection device 3 can detect the vehicle with higher accuracy.

Hereinafter, modified examples of the adjustment support device 2 and the vehicle detection device 3 of the embodiment will be described.
The vehicle detection device 3 may be configured to read the license plate based on the priority order set for each type of vehicle. Specifically, the storage unit 32 of the vehicle detection device 3 stores a priority table 321 that holds this priority.

FIG. 20 is a diagram showing a specific example of the priority order table 321.
The priority order table 321 has a priority order record for each gaze area. The priority order record includes gaze area ID, resolution, correction angle, and vehicle type 1 to vehicle type n. The gaze area ID is identification information of the gaze area. The resolution represents the number of pixels constituting the gaze area. That is, the resolution increases as the gaze area where the position close to the vehicle detection device 3 is imaged. The correction angle represents an angle of geometric correction performed on the gaze area. That is, the correction angle becomes larger as the gaze area where the position close to the vehicle detection device 3 is imaged. The resolution and the correction angle are acquired from the setting of the gaze area indicated by the gaze amount ID.

  Vehicle types 1 to n (n is an integer equal to or greater than 1) represent vehicle types classified according to vehicle characteristics such as size and height. This classification can be performed according to the type of pattern image that matches the image of the gaze area. The value of the priority set in each gaze area is registered in the values of the vehicle types 1 to n.

  The priority for each gaze area is determined by the operator of the adjustment work based on the detection result of the vehicle detection device 3. In this case, the vehicle detection device 3 accumulates information related to the detection result in the storage unit 32 of the own device, and transmits the information to the adjustment support device 2 when adjusting the own device. The operator confirms the detection result using the adjustment support device 2 and determines the priority order of the gaze area suitable for reading the license plate. The operator sets the determined priority order in the vehicle detection device 3.

FIG. 21 is a diagram illustrating a specific example of a screen for confirming the license plate recognition result.
The screen 800 is displayed on the display unit 22 of the adjustment support apparatus 2. The screen 800 includes, for example, an image display area 801, a number image display area 802, and a number information display area 803. The image display area 801 is an area in which an image from which the license plate has been read is displayed. The number image display area 802 is an area in which an image of a license plate detected from the image displayed in the image display area 801 is displayed. The number information display area 803 is an area in which information read from the license plate is displayed. The operator visually confirms the screen 800 and evaluates the detection result.

  Thus, the priority order of the gaze area determined based on the detection result of the vehicle detection device 3 is set for each type of vehicle, so that the vehicle detection device 3 can recognize the license plate more accurately. It becomes possible.

  Further, the vehicle detection device 3 may be configured to read the license plate based on position information set for each type of vehicle. Specifically, the storage unit 32 of the vehicle detection device 3 stores a position information table 322 that holds this position information.

FIG. 22 is a diagram illustrating a specific example of the position information table 322.
The position information table 322 has a position information record for each vehicle type. A position information record has each value of a vehicle type, start position information, end position information, and a correction angle. The vehicle type represents the type of vehicle classified according to the vehicle characteristics such as size and height. This classification can be performed according to the type of pattern image that matches the image of the gaze area. The start position information represents the position of the start point of the area where the license plate is detected. The end position information represents the position of the end point of the area where the license plate is detected. The correction angle represents the inclination of the area where the license plate is detected. The shape of the area where the license plate is detected is determined by the start position information, the end position information, and the correction angle.

  The position information for each vehicle type is determined by the operator of the adjustment work based on the detection result of the vehicle detection device 3. An operator confirms a detection result using the adjustment support apparatus 2, and determines the position where a license plate may be installed. The worker sets the determined position information for each vehicle type in the vehicle detection device 3.

  Thus, the position information determined based on the detection result of the vehicle detection device 3 is set for each type of vehicle, so that the vehicle detection device 3 detects a license plate installed at a position different from the normal position. It becomes possible to do.

  According to at least one of the embodiments described above, the accuracy of vehicle detection in the vehicle detection system includes the gaze area correction unit that corrects the image of the gaze area and the adjustment support apparatus that supports the installation or adjustment of the vehicle detection apparatus. Can be improved.

  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 their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 1 ... Vehicle detection system, 2 ... Adjustment support apparatus, 21 ... Communication part, 22 ... Display part, 23 ... Input part, 24 ... Memory | storage part, 25 ... Information acquisition part, 26 ... Adjustment information acquisition part, 27 ... Setting part, 271 ... Installation information setting unit, 272 ... Road region setting unit, 273 ... Image feature setting unit, 274 ... Gaze region setting unit, 3 ... Vehicle detection device, 31 ... Communication unit, 32 ... Storage unit, 321 ... Priority order table, 322 ... Position information table, 33 ... Imaging unit, 34 ... Vehicle candidate detection unit, 35 ... Gaze area correction unit, 36 ... Vehicle detection unit, 37 ... Vehicle state determination unit, 38 ... Vehicle counting unit, 4 ... Host system, 5 ... vehicle, 6 ... structure, 200 ... image, 201 ... road area, 202, 203 ... non-road area, 300 ... image, 400 ... image, 401, 402 ... gaze area, 501 to 503 ... gaze area, 601 and 602 ... imaging range, 03 ... vehicle, 701 ... image, 703 to 705 ... attentive area, 800 ... screen, 801 ... image display area, 802 ... Number image display area, 803 ... number information display area, V1 to V7 ... watch area

Claims (13)

An imaging unit that captures a road and obtains an image of the road;
For a plurality of gaze regions in the image where the vehicle traveling on the road is imaged, a geometric conversion unit that performs geometric conversion according to the installation position and orientation of the imaging unit;
A vehicle detection unit for detecting the vehicle from an image of a gaze area geometrically transformed by the geometry transformation unit;
A vehicle detection device comprising: The vehicle detection unit detects a front portion of the vehicle including a license plate from the image of the gaze area,
A state determination unit that determines the passage of the vehicle on the road based on the transition of the detection result of the vehicle in a plurality of gaze regions;
A counting unit that counts vehicles that have passed through the road based on the determination; and
The vehicle detection device according to claim 1. The vehicle detection unit reads the number of the vehicle from the image of the gaze area where the vehicle is detected.
The vehicle detection device according to any one of claims 1 and 2. The vehicle detection unit selects a gaze area from which the number of the detected vehicle is read based on the relationship between the priority order of the gaze area set in advance and the vehicle type.
The vehicle detection device according to claim 3. When the vehicle detection unit cannot read the number from the image of the gaze area in which the vehicle is detected, the vehicle detection unit reads the number from the region at a predetermined position in the preset image.
The vehicle detection device according to claim 3. An imaging unit that captures a road and obtains an image of the road;
A geometric conversion unit that performs geometric conversion according to an installation position and orientation of the imaging unit for a plurality of gaze regions in the image in which a vehicle traveling on the road is captured;
A vehicle detection unit for detecting the vehicle from an image of a gaze area geometrically transformed by the geometry transformation unit;
An adjustment support device for setting the vehicle detection device,
An installation information setting unit for setting information regarding the installation of the vehicle detection device;
A road area setting unit for setting an area where the road imaged in the image is located;
A feature setting unit for setting information on the position and shape of a subject that is constantly imaged in the image;
A gaze area setting unit for setting the gaze area;
An adjustment support apparatus comprising: The gaze area setting unit sets a plurality of gaze areas at intervals equal to or longer than the vehicle body length of the vehicle based on the area where the captured road is located.
The adjustment support device according to claim 6. The road area setting unit sets a position for each lane for a road composed of a plurality of lanes,
The gaze area setting unit sets a gaze area for each lane, masks an area in the gaze area where the same vehicle may be imaged in duplicate, and masks an area masked from the gaze area of one lane exclude,
The adjustment support device according to claim 7. The gaze area setting unit sets the gaze area where the license plate is not likely to be imaged and the vehicle type where the license plate is not likely to be imaged in the gaze area,
The adjustment support device according to claim 8. The gaze area setting unit sets priority of gaze areas determined for each vehicle type based on a detection result in the vehicle detection device;
The adjustment support device according to claim 9. The gaze area setting unit sets position information of a gaze area determined for each vehicle type based on a detection result in the vehicle detection device;
The adjustment support device according to claim 10. An imaging step of capturing a road and obtaining an image of the road;
A geometric conversion step for performing a geometric conversion according to an installation position and orientation of an imaging unit that acquires the image for a plurality of gaze regions in the image in which a vehicle traveling on the road is captured;
A vehicle detection step of detecting the vehicle from an image of the gaze area that has been geometrically transformed in the geometric transformation step;
A vehicle detection method comprising: An imaging unit that captures a road and obtains an image of the road;
A geometric conversion unit that performs geometric conversion according to an installation position and orientation of the imaging unit for a plurality of gaze regions in the image in which a vehicle traveling on the road is captured;
A vehicle detection unit for detecting the vehicle from an image of a gaze area geometrically transformed by the geometry transformation unit;
An adjustment method performed by an adjustment support device that communicates with a vehicle detection device including the setting of the vehicle detection device,
An installation information setting step for setting information regarding the installation of the vehicle detection device;
A road area setting step for setting an area where the road imaged in the image is located;
A feature setting step for setting information on the position and shape of a subject that is regularly imaged in the image;
A gaze area setting step for setting the gaze area;
An adjustment method comprising:

Images