JP4947592B2 - Vehicle detection device - Google Patents

Description

  The present invention relates to a vehicle detection device that analyzes an image obtained by photographing a vehicle traveling on a road with a camera and detects the vehicle separately from road surface reflected light.

  In an apparatus that captures road conditions with a camera, detects a vehicle from the image, and measures traffic, the vehicle detection at night is performed based on a high-luminance part such as a headlight. At this time, it is necessary not to misidentify the light part and the road surface reflected light, and various techniques for removing the road surface reflected light from the image have been proposed as exemplified below.

(1) A road surface reflected image is a technology that removes a reflected image by utilizing the three conditions of “vertical> horizontal, luminance decreases as it goes to the tip, and two reflected images are arranged” (for example, (See Patent Document 1). In this technology, when the aspect ratio of the reflected image is about the same as that of the light part in the vicinity of the camera, when the reflected image and the light part stick together, or when the reflected image is saturated, the light part Indistinguishable from reflection image.

(2) A technique for removing reflected light by utilizing the fact that the positional relationship of the light pair in the vertical and horizontal directions is the same position, the vertical distance is in the range, and the shape is similar (see, for example, Patent Document 2). In this technology, when the reflected light and the shape of the light are different, for example, the four high-intensity parts that are in the vertical / left / right positional relationship are “headlights and their reflected light” or “lamps and headlights at high positions in large vehicles”. I can't tell you.

(3) When the camera is installed so that the vehicle is photographed from diagonally forward, the amount of light incident on the camera is greater in the distance from the camera. Therefore, a technique for removing reflected light by increasing the threshold (for example, , See Patent Document 3). This technology cannot be applied when the brightness of light or reflected light is saturated (saturated), whether it is far or near the camera.

(4) A technique for removing a high-luminance portion having a long vertical length as road surface reflection (see, for example, Patent Document 4). With this technology, when the reflected light becomes small near the camera and becomes the same size as the light, it cannot be distinguished whether it is reflected light or light.

JP-A-3-286398 JP 11-353580 A JP-A-9-282448 JP-A-5-189694

  As illustrated in FIG. 7, the road surface reflected light R of the headlight H is photographed in a vertically long shape at a distance from the camera ((a) in the same figure), and slightly shortened at a middle distance ((b) in the same figure). In the vicinity ((c) in the figure), there is a property that it is small or not photographed. The techniques (1) and (4) take advantage of this property to remove road surface reflected light from distant images.

  When measuring the number of passing vehicles, if the measurement position is set to one place at a distance or a medium distance, the vehicle can be identified by removing reflected light with a low misperception rate using the above-mentioned properties. .

  However, when measuring the speed of the vehicle, it is necessary to track the vehicle, and it is necessary to track the vehicle. In particular, it is necessary to track the same portion of the vehicle in order to perform accurate speed measurement. As described above, since the shape of the road surface reflected light changes depending on the distance from the camera, it is difficult to track the same portion. Therefore, in order to correctly measure the vehicle speed, it is effective to track the light portion where the change in shape on the image due to the distance from the camera is small. Further, in order to increase the accuracy of speed measurement, it is desirable to set the vicinity of the camera where the movement amount in the image is larger than the far distance for the same movement distance in the real world as the vehicle speed measurement region.

For this purpose, it is necessary to determine whether the detected vehicle candidate area (high luminance part) is the road surface reflected light R or the light part H in the vicinity of the camera (such as the measurement position M in FIG. 7C). . However, when the vehicle travels from a distance of the camera toward the vicinity, the vehicle candidate area is correctly identified as the road surface reflected light or the light portion only from the information in the measurement area in the vicinity of the camera. It ’s difficult. The reason is as follows.
1. The state of road surface reflected light is different for each vehicle.
2. The shape of the road surface reflected light changes depending on the positional relationship between the vehicle and the camera.
3. It is difficult to distinguish between “road surface reflected light and light” or “light and other light emitters”.

  3. above. An example of this is shown in FIG. FIG. 6A shows an example of an image in which “light H and other light emitters G” are photographed at night. For example, various lamps (other light emitters G) are often installed on the top of a large truck. FIG. 4B is an example of an image in which “road surface reflected light R and light H” are photographed at night. In both images, when a high-intensity part is extracted, four points are extracted, but only from the data of these four points, “road surface reflected light R and light H” or “light H and other light emission”. It is difficult to distinguish whether it is “body G”.

  The present invention is intended to solve the above-mentioned problem, and accurately determines whether a vehicle candidate area in the vicinity of the camera on the image is a vehicle (particularly a light portion of the vehicle) or its road surface reflected light. An object of the present invention is to provide a vehicle detection device capable of detecting

  The gist of the present invention for achieving the object lies in the inventions of the following items.

[1] A camera that captures a passing vehicle from the front or rear thereof;
A candidate vehicle area having the characteristics of the vehicle is extracted from the image obtained by photographing with the camera, and the movement thereof is tracked, and whether or not the candidate vehicle area is road surface reflected light at one or a plurality of locations being tracked. Make a determination and store a history of the determination result, and determine whether the vehicle candidate region at the predetermined measurement position in the image is a vehicle or road surface reflected light based on the determination result history for the vehicle candidate region An analysis unit to perform,
I have a,
The analysis unit determines whether the vehicle candidate area at the measurement position is a vehicle or road surface reflected light. The vehicle candidate area at a middle distance or more away from the measurement position is easily distinguishable between the road surface reflected light and the light part. The vehicle detection apparatus characterized by determining based on the said determination result with respect to .

  In the above invention, the movement of the vehicle candidate area is tracked, and whether the vehicle candidate area is the vehicle or the road surface reflected light is determined during the tracking, and the history of the determination result is stored. Whether or not the vehicle is a vehicle is determined based on a history of determination results for the vehicle candidate area. For example, if the number of times that the vehicle candidate area at the measurement position is determined to be road surface reflected light during the tracking is equal to or greater than the prescribed number, the vehicle candidate area at the measurement position is determined to be road surface reflected light.

In the above invention, in the distant camera, since the vehicle (in particular light unit) and a road surface reflected light identifies from the vicinity of the camera easily, the determination for the vehicle candidate region in the vicinity of position, the distinction between the road surface reflected light and the light section This is based on a determination result at a middle distance or more that is likely to be attached and a distance from the measurement position .

[ 2 ] If the vehicle candidate region at the measurement position is determined to be road surface reflected light in the determination based on the determination result history, whether or not another vehicle candidate region exists in the vicinity of the far side of the vehicle candidate region. The vehicle detection device according to [1], wherein the vehicle candidate region is determined to be a vehicle if it exists, and if not, the vehicle candidate region at the measurement position is determined to be a vehicle.

  In the above invention, since the light part that is the light source should exist above the road surface reflected light (distant side), it corresponds to the light part near the upper part (far side) of the vehicle candidate area determined as the road surface reflected light. If the vehicle candidate area exists, it is determined as the light part (vehicle).

[ 3 ] The analysis unit accumulates and stores tracking data in which the position of the vehicle candidate area being tracked is associated with the time when the vehicle is located, and finally the vehicle is determined by the determination at the measurement position. The vehicle detection device according to [1] or [2] , wherein the moving speed of the vehicle is calculated from the tracking data for the determined vehicle candidate region.

[ 4 ] The vehicle detection device according to any one of [1] to [ 3 ], wherein the vehicle candidate region is a high-intensity portion having characteristics of a light portion of the vehicle.

  The vehicle detection apparatus according to the present invention can detect a vehicle with high accuracy without being affected by road surface reflected light even in a vehicle candidate region located in the vicinity of the camera on the image. This makes it possible to accurately measure the speed of the vehicle.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a schematic functional configuration of a vehicle detection apparatus 10 according to the present invention. The vehicle detection device 10 is a device for measuring traffic volume and measuring the speed of a passing vehicle by analyzing an image obtained by photographing a road condition with the camera 11, and an image obtained by photographing with the camera 11. The analysis part 12 which processes is provided. The camera 11 is fixed to a post or the like installed across the road or standing on the side of the road, and photographs a vehicle traveling on the road from the front, preferably obliquely from above, as shown in FIGS. It is installed to capture such images. In the present embodiment, a case will be described in which the camera 11 is installed obliquely in front of the vehicle and the vehicle travels from the far side to the vicinity of the camera 11.

  The analysis unit 12 includes an image acquisition unit 13, an image memory 14, a vehicle extraction unit 15, a road surface reflected light determination unit 16, a vehicle tracking unit 17, a speed measurement unit 18, and a tracking history management unit 19. Configured.

  The camera 11 captures an image of several frames to several tens of frames per second, and the image acquisition unit 13 captures the image data from the camera 11. Here, the camera 11 captures images every several tens of ms to 100 ms, and the angle of view is set so that a vehicle passing at a speed limit passes within the capturing range of the camera 11 in a few seconds. The image memory 14 temporarily stores the image data captured from the camera 11 by the image acquisition unit 13. The vehicle extraction unit 15 extracts a vehicle candidate area where a vehicle or a part of the vehicle exists by extracting a feature amount from the image. The feature amount is extracted in the daytime, for example, by extracting the contour line of the vehicle body. In addition, when the photographed image has relatively low brightness and the feature amount of the vehicle body cannot be acquired (for example, at night), the high brightness portion is acquired as the feature amount of the vehicle.

  In the following description, a case will be described in which a high-luminance portion having the characteristics of the vehicle, that is, a portion that seems to be a light portion of the vehicle is extracted as the vehicle candidate region 41 (see FIG. 7). The feature amount extraction is not limited as long as it is a process that can extract the feature amount of the vehicle. For example, in the case of nighttime, the vehicle candidate area 41 may be extracted based on an arbitrary element in the luminance and the shape, size, arrangement (such as left and right pairs) of the high luminance portion.

  The vehicle tracking unit 17 tracks the movement of the vehicle candidate area 41 extracted by the vehicle extracting unit 15, and the tracking history management unit 19 has a function of storing and managing the tracking history. The road surface reflected light determination unit 16 determines whether the vehicle candidate area 41 is a vehicle or road surface reflected light. This determination is performed sequentially (for each frame) during tracking, and the determination result is stored in the tracking history management unit 19 as part of the tracking history. In the case of nighttime, the vehicle candidate area 41 is a high-luminance part having a characteristic like a headlight. Therefore, the road-surface reflected light determination unit 16 determines whether the high-luminance part is a light part or road surface reflected light.

  The speed measurement unit 18 stores the vehicle speed of the vehicle candidate area 41 that has finally reached the measurement line M (see FIG. 7C) and is determined to be a vehicle in the tracking history management unit 19. It fulfills the function of calculating based on the tracking history.

  Note that the vehicle detection device 10 is actually composed of hardware as shown in FIG. That is, the analysis unit 12 includes a CPU (Central Processing Unit) 21, a ROM (Read Only Memory) 22, a RAM (Random Access Memory) 23, a camera I / F 24 to which the camera 11 is connected, an operation unit 25, a display unit 26, It is configured by connecting an external communication unit 27 and the like.

  The CPU 21 performs overall control of the operation of the vehicle detection device 10 by executing a program stored in the ROM 22. The ROM 22 stores various fixed data in addition to the above programs. The RAM 23 is used as a work memory for temporarily storing data when the CPU 21 executes a program, an image memory for storing image data captured from the camera 11, and a memory for storing a tracking history.

  The operation unit 25 has a function of accepting various setting operations (for example, an angle of view setting of the camera 11) with respect to the vehicle detection device 10, and includes a numeric keypad and a setting switch. The display unit 26 is configured by a liquid crystal display or the like, and fulfills a function of displaying a determination result, an operation state of the apparatus, setting contents, and the like. The external communication unit 27 functions to transmit determination results, speed measurement results, and the like to an external device and to receive remote control instructions from the outside.

  Next, the operation of the vehicle detection device 10 will be described.

  As illustrated in FIG. 7C, the vehicle detection device 10 includes a vehicle candidate region (high luminance portion) 41 that has reached a measurement line M set at a predetermined position in the vicinity of the camera 11 in the image. H) or road surface reflected light R is determined. At this time, it is difficult to distinguish whether the candidate vehicle region (high luminance part) 41 is the light part H of the vehicle or the road surface reflected light R only from the image of the candidate vehicle region (high luminance part) 41 that has reached the measurement line M. On the other hand, as shown in FIG. 7A, when the vehicle candidate region (high luminance part) 41 is far from the camera 11, the road surface reflected light R of the light part H is photographed vertically and the light part H is a light. Since the image is taken with the shape of the main part, it is easy to distinguish the road surface reflected light R from the light part H.

  Therefore, the movement of the vehicle candidate area (high brightness portion) 41 is tracked from the distance of the camera 11 until it reaches the nearby measurement line M, and the vehicle candidate area 41 is reflected by the road surface reflected light R at one or a plurality of locations during the tracking. Whether the vehicle candidate area 41 that has reached the measurement line M near the camera 11 is the light portion H of the vehicle or the road surface reflected light R is determined. This is performed based on a history of determination results (especially determination results at a distant location) during tracking for the vehicle candidate region (high luminance portion) 41. Then, the speed of the vehicle is calculated from the movement locus (tracking history) of the vehicle candidate area (high luminance part) 41 that is finally determined to be the light part H of the vehicle, and the speed is measured. .

FIG. 3 shows the flow of the entire vehicle detection process performed by the vehicle detection device 10.
First, the analysis unit 12 captures one frame of an image captured by the camera 11 (step S101), and extracts a vehicle candidate region (high luminance portion) 41 in the image (step S102). Then, a vehicle tracking process is performed for each extracted vehicle candidate area (high luminance part) 41 (step S103). In the vehicle tracking process, the movement of each vehicle candidate region (high luminance portion) 41 is tracked, and whether each vehicle candidate region (high luminance portion) 41 being tracked is a light portion H of the vehicle or road surface reflected light R is determined. The determination result is stored as a part of the tracking history.

  It is checked whether or not there is a vehicle candidate region (high luminance portion) 41 that has reached the measurement line M in the vehicle candidate region (high luminance portion) 41 extracted from the current image (step S104). If there is a vehicle candidate area (high luminance part) 41 that has arrived (step S104; Yes), speed calculation is performed for the vehicle candidate area (high luminance part) 41 (step S105). The above processing is repeated every time an image is captured at intervals of several tens of ms to 100 ms.

  FIG. 4 shows details of the vehicle tracking process (step S103 in FIG. 3). FIG. 5 shows an example of the data structure of the vehicle tracking data 50 accumulated and stored as a history during vehicle tracking. The vehicle tracking data 50 is managed as separate data for each vehicle candidate area (high luminance part) 41. FIG. 5 illustrates vehicle tracking data 50 of vehicle candidate numbers 51 “N−1”, “N”, and “N + 1”.

  Each vehicle tracking data 50 includes a vehicle candidate number 51 that is an identifier assigned to the vehicle candidate area (high luminance part) 41 being tracked, and the vehicle tracking process for the vehicle candidate area (high luminance part) 41. The tracking number counter 52 indicating the number of times performed, the coordinate history 53 indicating the coordinate position in the image of the vehicle candidate area (high brightness portion) 41, the time history 54 indicating the time when the vehicle is at the coordinate position, and the coordinate position Whether or not the speed measurement process has already been performed on the road surface reflection determination history 55 indicating the result of the determination of whether the road surface reflected light R or the light portion H of the vehicle has been performed, and the vehicle candidate region (high brightness portion) 41 The speed measured flag 56 shown in FIG. Among these, the coordinate history 53, the time history 54, and the road surface reflection determination history 55 are accumulated and stored for the number of tracking times, and the coordinate history 53, the time history 54, and the road surface reflection determination history 55 are associated with each time.

  When there is a vehicle candidate area (high luminance part) 41 newly detected by the vehicle detection process shown in step S102 of FIG. 3, new vehicle tracking data is added to the vehicle candidate area (high luminance part) 41. 50 is created and registered (FIG. 4, step S201). The processing from step S202 to step S206 is repeatedly performed for the number of times of the vehicle candidate area (high brightness portion) 41 currently being tracked (that is, the number of times of registered vehicle tracking data 50).

  In the process (steps S202 to S206) for each vehicle candidate area (high luminance part) 41 being tracked, the latest coordinate history 53 registered in the vehicle tracking data 50 of the vehicle candidate area (high luminance part) 41 is stored. Whether or not there is a vehicle candidate area (high luminance part) 41 extracted from the current image frame in the vicinity of the indicated coordinate position (the coordinate position of the vehicle candidate area (high luminance part) 41 in the previous frame) The vehicle candidate area (high luminance part) 41 is tracked by determining whether or not. The tracking of the vehicle candidate area (high luminance part) 41 in the vicinity area is performed by a method such as a method of determining based on the similarity of image patterns or a method of simply considering objects having close coordinates as the same object.

  When the vehicle candidate area (high luminance part) 41 corresponding to the vehicle candidate area (high luminance part) 41 being tracked (the vehicle candidate area (high luminance part) 41 of the previous frame) cannot be found in the current image frame (Step S203; No), the tracking for the vehicle candidate region (high luminance part) 41 is terminated at that time, and the vehicle tracking data 50 is discarded (Step S204). This occurs, for example, when a traveling vehicle goes out of the shooting range from the shooting range.

  On the other hand, when the vehicle candidate area (high luminance part) 41 corresponding to the vehicle candidate area (high luminance part) 41 being tracked can be found from the current image frame (the tracking vehicle is found), the discovered vehicle It is determined whether the candidate area (high luminance part) 41 is road surface reflected light R (step S205). Then, the tracking history (the current coordinate history 53 and the time history 54) that associates the determination result with the coordinate position of the found vehicle candidate area (high luminance part) 41 and the time when the current image frame was captured. (Tracking data associated with the road surface reflection determination history 55) is additionally registered in the corresponding vehicle tracking data 50 (step S206).

  Here, whether or not the light is road surface reflected light R is determined based on whether the aspect ratio of the shape of the vehicle candidate region (high brightness portion) 41 to be determined is higher or lower than a predetermined threshold. If the aspect ratio is longer than the threshold, the vehicle candidate area (high luminance part) 41 (tracking object) is determined as the road surface reflected light R, and the determination result is stored in the corresponding vehicle tracking data 50.

  At this time, whether or not the vehicle candidate region (high luminance portion) 41 having the characteristics of the light portion H is detected in the vicinity of the upper portion (far side) of the vehicle candidate region (high luminance portion) 41 determined to be road surface reflected light. If it is not detected, it is determined that the light portion H and the road surface reflected light R are integrated, and the upper end of the vehicle candidate region (high brightness portion) 41 determined to be the road surface reflected light R. The tracking process is performed with the range corresponding to the write part as the position where the write part H is located. That is, the determination result regarding the integrated vehicle candidate region (high luminance portion) 41 is changed to “not road surface reflected light”, and the position of the upper end portion is stored in the vehicle tracking data 50 as the coordinate position. In addition, the upper part (far side) vicinity assumes the maximum value which can be taken as the space | interval of the light part H and the road surface reflected light R, and means the range.

  FIG. 9 shows an example of an image when the light portion H and the road surface reflected light R are integrated. The upper part in the figure is the light part H, and the lower part is the road surface reflected light R. This is because the high luminance part is integrated due to the influence of blooming of the camera 11 or the like.

  FIG. 6 shows details of the speed calculation process (step S105 in FIG. 3). When the tracking target object (vehicle candidate area (high luminance part) 41) being tracked exceeds the measurement line M, the tracking history (vehicle tracking data 50) relating to the vehicle candidate area (high luminance part) 41 is used for speed. Calculate

  Specifically, it is determined whether or not the vehicle candidate region (high brightness portion) 41 that exceeds the measurement line M is the road surface reflected light R (step S301). Whether or not it is the road surface reflected light R refers to the tracking history (vehicle tracking data 50) of the candidate vehicle region (high luminance portion) 41 retrospectively to a predetermined position far from the camera 11. Although it may be set as appropriate as far as it goes back, it is necessary to include at least a distant position where the light portion H and the road surface reflected light R can be easily determined. The tracking history in the vicinity of the measurement line M is excluded from the count target, and for example, the number of times that the road surface reflected light R is determined is counted with reference to the tracking history only for a range farther from the middle distance. May be.

  Further, even in the vehicle candidate area (high luminance part) 41 that exceeds the measurement line M, the speed measured flag 56 of the vehicle tracking data 50 relating to the vehicle candidate area (high luminance part) 41 is already set. If this is the case, the speed calculation process will not be performed.

  If the number of times that it is determined as the road surface reflected light R in the tracking history is equal to or greater than a predetermined number of times, it is determined that the vehicle candidate region (high brightness portion) 41 is the road surface reflected light R. When it is determined by the determination based on the tracking history that the vehicle candidate region (high luminance portion) 41 is not the road surface reflected light R (step S301; No), it is determined as the light portion H of the vehicle, and the vehicle candidate region ( The speed is calculated based on the tracking history (vehicle tracking data 50) of the high luminance part) 41, and the speed measured flag 56 of the vehicle tracking data 50 is set (step S303).

  The predetermined number of times used as the determination criterion may be, for example, one time or may be two to several times. Furthermore, instead of the number of times, the road surface reflected light R is determined with respect to the number of times of tracking history to be referred to retroactively. It may be determined by the ratio of the number of times.

  When it is determined that the vehicle candidate area (high luminance part) 41 that has reached the measurement line M is the road surface reflected light R from the tracking history (step S301; Yes), the vehicle candidate area (high luminance part) on the screen. It is checked whether or not a vehicle candidate area (high luminance part) 41 corresponding to the light part H exists near the upper part (distant side) of 41 (step S302). When the candidate vehicle area (high luminance part) 41 corresponding to the light part H does not exist near the upper part (step S302; No), the vehicle candidate area (high luminance part) 41 that has reached the measurement line M is tracked earlier. It is determined that the determination result of the road surface reflected light R based on the history is incorrect, and the vehicle candidate area (high luminance part) 41 is determined as the light part H of the vehicle. Then, the moving speed of the vehicle is calculated from the tracking history (vehicle tracking data 50) of the vehicle candidate area (high luminance part) 41 determined to be the light part H, and the speed of the vehicle tracking data 50 has been measured. The flag 56 is set (step S303).

  When the vehicle candidate area (high brightness part) 41 corresponding to the light part H is found near the upper part (step S302; Yes), the discovered vehicle candidate area (high brightness part) 41 is used as the light part H of the vehicle. Is finally determined, the speed is calculated from the tracking history (vehicle tracking data 50) of the found vehicle candidate area (high luminance portion) 41, and the speed measured flag 56 of the vehicle tracking data 50 is set. (Step S304).

  Thus, since the light source should exist near the upper part of the road surface reflected light R, the vehicle candidate area 41 in the vicinity of the upper part is determined as the light part H with reference to the road surface reflected light R. Accuracy is improved. In addition, by setting the speed measurement completed flag 56, when this vehicle candidate area (high luminance part) 41 subsequently reaches the measurement line M, execution of the speed calculation process again is prevented. Furthermore, since the vehicle detection device 10 is configured to measure the traffic volume of the vehicle by counting the number of candidate vehicle regions (high brightness portions) 41 that have been subjected to the speed calculation process, the speed measured flag 56 By providing, double counting is prevented and traffic volume can be measured correctly.

  In calculating the speed, distance data in the camera image (data for associating the distance in the camera image with the actual distance) set in advance is used.

  Further, as shown in FIG. 7B, the vehicle candidate region (high luminance portion) 41 in which the light portion H and the road surface reflected light R are integrated as shown in FIG. In the case where the vehicle is separated vertically, the upper end portion of the vehicle candidate area (high brightness portion) 41 that has been integrated before separation is set as the reference position for tracking, so that the upper portion after separation is positioned relative to the object before separation. Close to. Therefore, the upper part after separation is continuously tracked as an object before separation. On the other hand, the lower part after the separation is recognized as a new vehicle candidate area (high luminance part) 41, a new vehicle candidate number 51 is assigned, and new vehicle tracking data 50 is provided, from which tracking is performed. Be started.

  As described above, in the vehicle detection device 10 according to the present embodiment, the vehicle candidate region (high brightness portion) 41 that has reached the measurement line M near the camera 11 is the light portion H of the vehicle or the road surface reflected light R. Is determined based on the determination result (tracking history) performed when the vehicle candidate area (high brightness portion) 41 is far from the measurement line M, and therefore only from the image near the measurement line M. Even when it is not possible to determine whether the light portion H or the road surface reflected light R is present, it is possible to accurately determine whether the vehicle candidate region (high luminance portion) 41 is the road surface reflected light R or the light portion H. And since the speed of the vehicle candidate area (high luminance part) 41 is calculated from the tracking history of the vehicle candidate area (high luminance part) 41 finally determined as the light part H, the part whose shape does not change on the image The speed can be calculated correctly based on the above.

  The embodiment of the present invention has been described with reference to the drawings. However, the specific configuration is not limited to that shown in the embodiment, and there are changes and additions within the scope of the present invention. Are also included in the present invention.

  In the embodiment, the case where the vehicle travels from the far side to the near side of the camera 11 has been described as an example. It is possible to increase the determination accuracy of H or road surface reflected light R.

  Moreover, although the vehicle candidate area | region 41 was demonstrated as a high-intensity part, the process sequence of this invention can be applied also to the analysis of the vehicle of the daytime as it is. That is, since the road surface reflected light R does not exist during the daytime, it is always determined that the vehicle candidate area 41 is not the road surface reflected light R, and the speed of the vehicle is measured.

  In the embodiment, an image is captured every tens of ms to 100 ms and a tracking history is left. However, as long as the movement of the vehicle can be tracked, the image capturing interval may be changed as appropriate. Note that the determination performed during tracking to the measurement line M is preferably performed every time an image is captured as illustrated in the embodiment, but it is easy to determine whether the light portion H or the road surface reflected light R is present. You may be comprised so that it may be performed only at least one place or several places of a distant side. In this case, the processing load of analysis is reduced.

It is a block diagram which shows schematic function structure of the vehicle detection apparatus concerning embodiment of this invention. It is a block diagram which shows the hardware structural example of the vehicle detection apparatus concerning embodiment of this invention. It is a flowchart which shows the flow of the whole vehicle detection process which the vehicle detection apparatus concerning embodiment of this invention performs. It is a flowchart which shows the detail of a vehicle tracking process (step S103 of FIG. 3). It is explanatory drawing which shows the example of a data structure of the data for vehicle tracking. It is a flowchart which shows the detail of a speed calculation process (step S105 of FIG. 3). It is explanatory drawing which illustrated the example of an image by which the light part of the vehicle and its road surface reflected light were image | photographed about three types, the position of a light part far from a camera, a middle distance, and the vicinity. It is explanatory drawing which illustrated the similarity with the case where the light part of a vehicle and the other light part provided in the upper part of the vehicle were image | photographed, and the case where the light part of a vehicle and its road surface reflected light were image | photographed. is there. It is explanatory drawing which shows the example of an image when the light part of a vehicle and its road surface reflected light are integrated by blooming etc.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Vehicle detection apparatus 11 ... Camera 12 ... Analysis part 13 ... Image acquisition part 14 ... Image memory 15 ... Vehicle extraction part 16 ... Road surface reflected light determination part 17 ... Vehicle tracking part 18 ... Speed measurement part 19 ... Tracking history management part 21 ... CPU
22 ... ROM
23 ... RAM
24 ... Camera I / F
25 ... Operation unit 26 ... Display unit 27 ... External communication unit 41 ... Vehicle candidate area (high luminance part)
DESCRIPTION OF SYMBOLS 50 ... Vehicle tracking data 51 ... Vehicle candidate number 52 ... Tracking number counter 53 ... Coordinate history 54 ... Time history 55 ... Road surface reflection determination history 56 ... Speed measurement completed flag H ... Light part R ... Road surface reflected light G ... Other light emission Body M ... Measurement line

Claims (4)

A camera that captures the vehicle in front or behind it,
A candidate vehicle area having the characteristics of the vehicle is extracted from the image obtained by photographing with the camera, and the movement thereof is tracked, and whether or not the candidate vehicle area is road surface reflected light at one or a plurality of locations being tracked. Make a determination and store a history of the determination result, and determine whether the vehicle candidate region at the predetermined measurement position in the image is a vehicle or road surface reflected light based on the determination result history for the vehicle candidate region An analysis unit to perform,
I have a,
The analysis unit determines whether the vehicle candidate area at the measurement position is a vehicle or road surface reflected light. The vehicle candidate area at a middle distance or more away from the measurement position is easily distinguishable between the road surface reflected light and the light part. The vehicle detection apparatus characterized by determining based on the said determination result with respect to . If the vehicle candidate area at the measurement position is determined to be road surface reflected light in the determination based on the determination result history, it is checked whether another vehicle candidate area exists near the far side of the vehicle candidate area, The vehicle detection device according to claim 1, wherein when the vehicle exists, the vehicle candidate region is determined as a vehicle, and when the vehicle candidate region does not exist, the vehicle candidate region at the measurement position is determined as a vehicle. The analysis unit accumulates and stores tracking data in which the position of the vehicle candidate area being tracked is associated with the time when the vehicle is located, and is finally determined to be a vehicle by the determination at the measurement position. vehicle detection apparatus according to claim 1 or 2, characterized in that to calculate the moving speed of the vehicle from the tracking data for the vehicle candidate region. The vehicle detection device according to any one of claims 1 to 3 , wherein the vehicle candidate area is a high-intensity part having characteristics of a light part of the vehicle.

Images