JP2002024810A - Traffic flow measuring device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately sense existence of a vehicle, without being influenced by a shadow of a building, a roadside tree or the like reflected on a road and a shadow of the moving vehicle, and to realize sensing at a low cost. SOLUTION: A vehicle extracting part 24 takes in image data from a camera 12 at prescribed time intervals, and takes out a normal direction time difference image changed from darkness to brightness and a reverse direction time difference image changed from brightness to darkness and executes expansion processing thereof. Normal and reverse time difference expanded images are added together, and subjected to binarization processing, and a logic product of past N times of the binary image is taken and used as a moving vehicle front part. Normalized Y-axis projection processing is executed, and binarization is executed, to thereby determine head coordinates of each moving vehicle. The head coordinates of each moving vehicle are tracked, and a vehicle control table 20 is renewed successively, and this-time coordinates vanish. When the preceding-time coordinates are in an outlet region, an correlation operation between a this-time outlet region image and an outlet region background image is executed. When there is no correlation, the vehicle information is reserved, and when there is correlation, the vehicle information is erased from the vehicle control table 20.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a traffic flow measuring device, and more particularly, to a traffic flow measuring device having image processing means for processing a bird's-eye view image from a camera installed at a high place on a road. The present invention relates to a traffic flow measurement device that measures a traffic flow by grasping the presence of a vehicle in a measurement area provided in a line direction.

[0002]

2. Description of the Related Art Conventionally, an image of a bird's-eye view of a vehicle coming and going is taken from a TV camera installed at a roadside high place such as a signal lamp pole, and the taken image is subjected to image processing to obtain a vehicle traffic direction of a lane. A traffic flow measurement device that measures the presence or the number of vehicles in a measurement area provided in a vehicle has been used. For example, in the “traffic information processing method” described in Japanese Patent Application Laid-Open No. H11-219493, a road image (background image) in a situation where no vehicle is present is collected in advance using a monocular camera, and a captured image at the time of measurement is taken. By comparing with, the presence of the vehicle was sensed. Also, Japanese Patent Application Laid-Open No. 10-1
In the "traffic flow monitoring device" described in Japanese Patent No. 54292, a vehicle is tracked by registering a feature portion of the front and rear surfaces of the vehicle as a template using a monocular camera and performing pattern matching processing while updating the template. Was. Further, the "traffic volume measurement device and the traffic signal control device" described in Japanese Patent Application Laid-Open No. H11-175883 uses a plurality of imaging devices that image the same direction on a road, and uses an image signal from the imaging device. The number of vehicles and the like is measured by stereo image processing for three-dimensional measurement.

[0003]

However, in such a conventional traffic flow measuring device, in the above case,
It is premised that an accurate background image is collected, but it is difficult to collect a stable background image on a road where vehicles frequently come and go, and image processing is performed using an image captured by a single-lens camera. For this reason, there is a problem that the vehicle is easily affected by the shadows of buildings and street trees reflected on the road, and the vehicle cannot be detected accurately. Also, in the above case, since a monocular camera is used, there is no means for removing the shadow in the feature amount of the vehicle to be registered in the template. There was a problem that can not be. Furthermore, in the above case, since stereo image processing is performed using a plurality of imaging devices, it is possible to deal with the influence of shadows and overlapping vehicles, and it is possible to accurately detect the presence of a vehicle. Installing an imaging device,
Since three-dimensional measurement is performed based on the obtained image signal, there is a problem that the cost is high. The present invention has been made in view of the above problems, and can accurately detect the presence of a vehicle without being affected by the shadows of buildings and street trees reflected on the road and the shadow of a moving vehicle, and at a low cost. It is an object of the present invention to provide a traffic flow measuring device that can be realized.

[0004]

According to a first aspect of the present invention, there is provided an image processing means for performing image processing of a bird's-eye view image from a camera installed at a high place on a road, provided in the direction of the traffic line of the lane. In the traffic flow measuring device which measures the traffic flow by grasping the existence state of the vehicle in the measured measurement area, the image processing means changes from dark to bright between two continuous road overhead images at predetermined time intervals. The forward time difference image extracting only the portion that has been changed and the reverse time difference image extracting only the portion that has changed from bright to dark are respectively expanded, and the moving vehicle is moved based on the image of the portion where the two expanded images overlap. Characterized by being provided with at least a vehicle extracting unit for extracting the front part of the vehicle. According to this, the image of the bird's-eye view image from the camera at the high point on the road is processed by the image processing means. The image processing means includes at least a vehicle extraction unit, and extracts a forward time difference image that extracts only a portion that changes from dark to bright between two continuous road overhead images at predetermined time intervals; The backward time difference image extracting only the changed part is expanded, and the front part of the moving vehicle is extracted based on the image of the part where the two expanded images overlap. For this reason, it is possible to remove not only shadows of fixed objects such as buildings and street trees reflected on the road, but also shadows of moving vehicles and regions having a uniform density such as general vehicles, and thus each moving vehicle can be removed. You can extract the front part of the vehicle to accurately grasp the presence of vehicles in the measurement area and measure traffic flow.
The cost can be reduced because it can be implemented by a monocular camera. According to a second aspect of the present invention, in the traffic flow measuring device according to the first aspect, the vehicle extracting unit further extracts an image of a portion where the two expanded images overlap with each other as a plurality of continuous images at a predetermined interval. The front part of the moving vehicle is extracted based on the result of the logical product of the plurality of images. According to this, the vehicle extracting unit included in the image processing means further extracts an image of a portion where the two expanded images overlap each other as a plurality of continuous images at a predetermined interval, and takes a logical product of the plurality of images. The front part of the moving vehicle is extracted based on the result. For this reason, for example, even if the shadow is divided by a trailer type truck having a gap between the driver's seat and the cargo bed portion, the shadow at the boundary portion can be removed.

According to a third aspect of the present invention, in the traffic flow measuring device according to the first aspect, the vehicle extractor corresponds to a pair of headlights or a vehicle width light when performing nighttime image processing. The front part of the moving vehicle is extracted based on the image. According to this, when performing nighttime image processing, the vehicle extracting unit included in the image processing unit extracts the front part of the moving vehicle based on an image corresponding to a pair of headlights or a vehicle width light. Therefore, during nighttime image processing, each moving vehicle is accurately extracted by recognizing the headlights and side lights of the moving vehicle that are lit at night, and the presence status of the vehicle in the measurement area is grasped. The traffic flow can be accurately measured. According to a fourth aspect of the present invention, in the traffic flow measuring device according to any one of the first to third aspects, vehicle information extracted from a measurement area of the road overhead view image is stored and managed for each vehicle. And the image processing means registers the vehicle information in the vehicle management table each time the vehicle extraction unit extracts a new moving vehicle front part in the measurement area, and The vehicle coordinates are updated while tracking the front of the vehicle, and when the front of the moving vehicle disappears in the middle of the measurement area, the vehicle is determined to be a stopped vehicle and the vehicle coordinates at that time are stored, and when the movement of the stopped vehicle is resumed, the stored vehicle coordinates are stored. And a vehicle tracking unit that updates the vehicle coordinates by restarting the tracking of the front of the moving vehicle. According to this, a vehicle management table for storing and managing vehicle information extracted from within the measurement area of the road overhead view image for each vehicle is provided. Then, in the vehicle tracking unit of the image processing means, each time the vehicle extraction unit extracts a new front of the moving vehicle in the measurement area, the vehicle information is registered in the vehicle management table, and the front of the moving vehicle is tracked. When the vehicle coordinates are updated, and the front of the moving vehicle disappears in the middle of the measurement area, it is determined that the vehicle is stopped and the vehicle coordinates at that time are stored.When the stopped vehicle resumes moving, the front of the moving vehicle is tracked from the stored vehicle coordinates. Restart and update vehicle coordinates. For this reason, it is possible to accurately track each moving vehicle in the measurement area while extracting the moving vehicles described above.

According to a fifth aspect of the present invention, in the traffic flow measuring device according to the fourth aspect, the image processing means provides an exit area near a vehicle exit in the measurement area, and the vehicle tracking unit controls a moving vehicle. If the front part disappears after entering the exit area, a correlation operation is performed between the background image of the exit area immediately before the front part of the moving vehicle enters the exit area and the image of the current exit area. If the calculated value is less than the predetermined threshold, the vehicle information is deleted from the vehicle management table. Is further provided in the vehicle management table. According to this, the image processing means further includes a vehicle passage determination unit. This vehicle passage determination unit provides an exit area near the vehicle exit of the measurement area, and when the vehicle tracking unit disappears after the front of the moving vehicle enters the exit area, immediately before the front of the moving vehicle enters the exit area. Performs a correlation operation between the background image of the exit area and the image of the current exit area, and deletes the vehicle information from the vehicle management table as a vehicle has passed if the calculation result exceeds a predetermined threshold, If the calculation result is equal to or less than the predetermined threshold value, it is determined that a part of the vehicle body remains, and the vehicle information is stored in the vehicle management table. Therefore, in the vehicle management table, the information of the moving vehicle existing in the measurement area is accurately registered and stored, and is deleted upon passing,
Just by looking at the vehicle management table, the presence state of the vehicle in the measurement area can be accurately grasped. According to a sixth aspect of the present invention, in the traffic flow measuring device according to the fourth aspect, the image processing means provides an exit area near a vehicle exit of the measurement area, and the vehicle extraction unit according to the third aspect operates at night. The vehicle at the time is extracted, and when the front part of the moving vehicle disappears after entering the exit area by the vehicle tracking unit, the exit area is set to 2 in brightness.
If the ratio of the high brightness area in the exit area is equal to or less than a predetermined threshold, the vehicle information is deleted from the vehicle management table assuming that the vehicle has passed, and if the ratio exceeds the predetermined threshold. For example, a vehicle passage determination unit that stores the vehicle information in the vehicle management table assuming that a part of the vehicle body remains is further provided. According to this, the vehicle extraction and vehicle tracking at night use the headlights and side lights of the moving vehicle, but the vehicle passage determination unit at nighttime uses the headlights of the vehicles to be determined. When the vehicle side light passes, it disappears from the screen, so whether or not a part of the vehicle body still remains in the exit area cannot be determined by the presence or absence of the headlight and the vehicle width light. For this reason, whether or not a part of the vehicle body remains in the exit area at night is determined by checking the presence or absence of side reflections caused by a speed light, a side light, or a headlight in another lane attached to the vehicle. When the proportion of the high brightness area is large, it is assumed that a part of the vehicle body remains in the exit area.

According to a seventh aspect of the present invention, there is provided the traffic flow measuring device according to any one of the fourth to sixth aspects, wherein the vehicle in the measurement area is based on the vehicle information stored in the vehicle management table. And a measurement result output unit that obtains at least one measurement result of the presence of the vehicle, the number of vehicles, or the number of passing vehicles, and outputs the result to the outside. According to this, the measurement result output unit:
By referring to the vehicle information stored in the vehicle management table, obtaining at least one measurement result of the presence, the number of vehicles, or the number of passing vehicles in the measurement area, and outputting the result to the outside, The above traffic flow measurement result can be used, for example, for traffic signal control, or provided to an on-vehicle device or the like via an optical beacon. According to an eighth aspect of the present invention, in the traffic flow measuring device according to any one of the first to seventh aspects, a road overhead view image from a camera installed at a high point on the road is a bird's-eye view of a front part of the moving vehicle. It is characterized in that image processing is performed using an overhead image of the rear surface of the moving vehicle instead of the image. According to this, the bird's-eye view image from the camera installed at a high place on the road may be not only an image that looks down at the front part of the moving vehicle, but also an image that looks down at the rear part of the moving vehicle. Also, by performing the same image processing as described above, the traffic flow can be accurately measured. According to a ninth aspect of the present invention, in the traffic flow measuring device according to any one of the first to seventh aspects, the road overhead view image from a camera installed at a high place on the road is a bird's eye view of the front part of the moving vehicle. The present invention is characterized in that an on-line simultaneous measurement is made possible by performing image processing by placing a traffic lane and a traffic lane that overlooks the rear part of the moving vehicle in the same image field of view. According to this, a road overhead view image from a camera installed at a high place on the road is an image in which a lane that looks down on the front part of the moving vehicle and a lane that looks down on the rear part of the moving vehicle are included in the same image field of view. The processing may be performed, and the simultaneous simultaneous measurement of the traffic flow becomes possible.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a traffic flow measuring device 10 according to the present embodiment. 1 includes a camera 12, an image input unit 14, an image storage unit 16, an image processing unit 18, a vehicle management table 20, a measurement result output unit 22, and the like. The image processing unit 18 further includes a vehicle extracting unit 24, a vehicle tracking unit 26, a vehicle passing determining unit 28, and the like. The camera 12
Here, a monocular TV camera using a solid-state image pickup device (CCD) on an image pickup surface is disposed at a roadside high place such as a lamp pole, and has an angle as shown in FIG. 2 or FIG. Is a bird's-eye view. The image input unit 14 includes an A / D conversion unit that converts analog image data input from the camera 12 into digital image data. The image storage unit 16 temporarily stores the image data that has been captured by the camera 12 and converted into digital data, and the operation of image processing in each unit of the image processing unit 18 described below and the operation of storing the result are repeated. .
The image processing unit 18 performs image processing using the digital image data stored in the image storage unit 16 so that the presence measurement area (a) 36 set in the captured image as shown in FIGS. This is for grasping the existence state of the moving vehicle in the exit area (b) 38. Note that the four radially extending lines 30, 32, 34, and 35 in FIGS.
This is a line showing three lanes on one side, and an opposite lane 40 (not shown) is on the left side. Specifically, image processing is performed in each of the following units. Vehicle extraction unit 24
Then, a captured image data as shown in FIG. 4 is fetched at predetermined time intervals, and a forward direction time difference image (see FIG. 5) for extracting only a portion changed from dark to bright between two continuous images, And an inverse time difference image (see FIG. 6) for extracting only a portion that has changed to a dark color. At this point, the shadow 48 of the fixed structure such as a building is removed. Then, the forward direction time difference image and the backward direction time difference image are subjected to expansion processing, added, and then binarized with a predetermined threshold value.
Areas with dense horizontal stripes as shown in (for example, front and rear parts of vehicles with grills, bumpers, etc.)
Can be extracted, and a region having a uniform density such as the shadow 44 of the moving vehicle 42 can be removed.

In the case of a trailer-type truck having a gap between the driver's cab and the loading platform as shown in FIG. May not be removed. In this case, a plurality of images (FIG. 8B, current image result, (c) previous image result, and (d) two-time previous image result) are successively imaged at predetermined time intervals with respect to the processing result shown in FIG. Storage unit 16
, And by taking the logical product of those images, it is possible to remove shadows and noise at the boundary. Further, when the vehicle is extracted at night, only a pair of headlights or a vehicle width light (such as at twilight) of the vehicle is picked up by the camera when the vehicle is extracted at night. And a vehicle width light is a front part (or a rear part) of the moving vehicle. In addition, since there is no shadow problem at night, even a trailer-type truck can extract the front of the moving vehicle using only the headlights and the side lights. Next, the vehicle tracking unit 26 tracks the front surface of the moving vehicle within the presence measurement area 36 set in the lanes 30, 32, 34, and 35. For example, every time the vehicle extraction unit 24 extracts the front of the moving vehicle in the presence measurement area 36, the extracted vehicle information is stored in the vehicle management table 20 (see FIG. 3).
Register a new one. At this time, the new registration flag in FIG.
N, the movement flag is set to ON, and the coordinates are registered this time. Further, the vehicle tracking unit 26 tracks the front of the moving vehicle using the vehicle extracting unit 24 and updates the vehicle management table of the vehicle. At this time, the new registration flag in FIG.
F, the movement flag is turned ON, the previously registered coordinates are moved to the previous coordinates, and the current coordinates are updated. When the moving vehicle stops in the presence measurement area 36 due to a signal or the like, the front surface of the moving vehicle disappears. Therefore, the vehicle tracking unit 26 determines that the vehicle is stopped and updates the vehicle management table 20 of the vehicle. I do. At this time, the new registration flag and the movement flag in FIG. When the stopped vehicle moves again, the vehicle management table 20 of the stopped vehicle is updated. At that time, FIG.
The new registration flag in the middle is OFF, the movement flag is ON,
Move the saved coordinates to the previous coordinates and update the coordinates this time.
Such tracking of the vehicle is performed for each moving vehicle (vehicle 1, vehicle 2,...) In the presence measurement area 36.

[0010] Next, when the front of the moving vehicle enters the exit area 38 and disappears during the daytime, the vehicle passage determination section 28 determines whether the entire vehicle has passed or a part of the vehicle body still remains. It is not clear whether it is in the state. Therefore,
In the angle of FIG. 4, the background image of the exit area stored immediately before the vehicle enters the exit area 38 and the current exit are utilized by utilizing the fact that the inter-vehicle resolution near the exit area 38 is high and the background image is easily available. A correlation operation with the image of the area 38 is performed. As a result, if the value exceeds a predetermined threshold value, it means that the correlation is high. Therefore, it is determined that the vehicle has passed, and the vehicle information is deleted from the vehicle management table 20. If the calculation result is equal to or less than the predetermined threshold value, it means that the correlation is low, so that it is determined that a part of the vehicle body remains, and the vehicle management table 20 of the vehicle is stored. In addition, the background image is when the front part of the moving vehicle comes just before and there is no vehicle body left in front of it,
In addition, updating is performed only when the three requirements when the correlation calculation result with the background image last time exceeds a predetermined threshold value are satisfied. Also, in the case where the front part of the moving vehicle disappears after entering the exit area at night, the vehicle passage determination unit 28 makes it clear whether the whole vehicle has passed or the part of the vehicle body remains as in the daytime. is not. Therefore, in the present embodiment, binarization based on the brightness of the exit area 38 is performed, and if a high brightness area is equal to or less than a predetermined threshold value outside the coordinate area indicated by the vehicle management table, the speed light, It is determined that the vehicle has passed without any side reflection or the like due to a lamp or a vehicle headlight in another lane, and the vehicle information is deleted from the vehicle management table 20. When the high brightness area exceeds a predetermined threshold, there is a speed light, a side light, or a side reflection by a vehicle headlight in another lane, so that if a part of the vehicle body remains. The determination is made, and the vehicle management table 20 of the vehicle is stored.

Next, the operation of this embodiment will be described. FIG. 9 is a flowchart illustrating a processing operation of the traffic flow measuring device according to the present embodiment. The traffic flow measurement device according to the present embodiment measures the traffic flow of a moving vehicle in the presence measurement area 36 shown in FIG. 4 based on image data obtained by capturing a three-lane vehicle road with a single-lens camera. It performs signal control and provides traffic information to an on-vehicle device or the like via an optical beacon. First, an input image from the camera 12 is processed as a current image (step S10).
0). Subsequently, in step S102, a set of positive values is obtained. That is, here, the vehicle extraction unit 24 captures captured image data at predetermined time intervals, and extracts a forward direction time difference image as shown in FIG. 5 in which a portion changed from dark to bright between two consecutive images is extracted. This is further expanded. In step S104, a set of negative values is obtained. That is, here, the vehicle extraction unit 24 captures the captured image data at a predetermined time interval, and extracts a reverse time difference image as shown in FIG. 6 in which a portion changed from bright to dark between two consecutive images is extracted. This is further expanded. Then, after adding the above-described forward / reverse time difference dilated images, a binarization process is performed (step S106). Note that, although different from the operation of the present embodiment, a similar result can be obtained by binarizing each of the forward / reverse time difference dilated images and then calculating the logical product of these two images. That is, here
A process of obtaining an image of an overlapping portion of the forward / reverse time difference expanded image is performed.

In the present embodiment, the logical product of the binary image obtained in step S106 for the past N times is calculated so as to be applicable to a trailer-type truck having a gap between the driver's seat and the bed portion. This is taken as the front of the moving vehicle (step S108). Then, after performing the normalized Y-axis projection process, binarization is performed again to obtain the leading coordinates for each moving vehicle (step S112). In step S114, the vehicle management table 20 is sequentially updated while tracking the leading coordinates of each moving vehicle. For example, new vehicle registration, moving vehicle coordinates update, stop vehicle coordinates storage, and the like are performed. Then, in step S116, it is determined whether or not the coordinates have disappeared this time.
Then, the process proceeds to 18, where it is determined whether or not the last time coordinates were in the exit area. Here, if the previous coordinate is the exit area,
In step S120, a correlation operation is performed between the current exit area image and the exit area background image, and in step S122, the presence or absence of a correlation is determined. If it is determined in step S122 that there is no correlation, since a part of the vehicle body remains in the exit area, the vehicle management table 20 is updated and the vehicle information is stored (step S124). If it is determined in step S116 that the coordinates have not disappeared this time, or if the previous coordinates are not in the exit area in step S118, the process proceeds to step S126. Further, when it is determined in step S122 that there is a correlation,
Since the passage of the vehicle has been confirmed, the vehicle management table 20 is updated, and the vehicle information is updated to the vehicle management table 20.
(Step S128). In step S126, the measurement result output unit 22 outputs a detection signal every time a newly registered vehicle starts up in the vehicle management table 20 to count the number of passing vehicles within a predetermined time. When there is a registered vehicle in the vehicle management table 20, the presence of the vehicle in the presence measurement area and the number of existing vehicles are output.

As described above, according to the present embodiment, the cost can be reduced by using a monocular camera, and image processing is performed using forward and reverse time difference images. It is no longer affected by the shadow of a tree or the like and the shadow of the moving vehicle, and the presence of the moving vehicle can be accurately detected. Therefore, by performing traffic signal control using the measurement result of the traffic flow measuring device of the present embodiment or providing traffic information to the vehicle-mounted device, control and information provision based on accurate actual information can be performed. In the above-described embodiment, the camera is installed at a position where the front part of the moving vehicle is imaged, and is implemented as a device that measures traffic flow in one lane. However, the present invention is not limited to this. It is also possible to perform the facing simultaneous measurement while looking down at both the part and the rear part. Further, in the above embodiment, the image processing unit 18 is configured by the vehicle extracting unit 24, the vehicle tracking unit 26, and the vehicle passing determining unit 28, but is not necessarily limited to this component. It is also possible to change the components according to the classification method and the processing procedure.

[0014]

As described above, according to the first aspect of the present invention, an image of a bird's-eye view image from a camera at a high point on the road is processed by the image processing means. The image processing means includes at least a vehicle extraction unit, and extracts a forward time difference image that extracts only a portion that changes from dark to bright between two continuous road overhead images at predetermined time intervals; The backward time difference image extracting only the changed part is expanded, and the front part of the moving vehicle is extracted based on the image of the part where the two expanded images overlap. For this reason,
In addition to removing the shadows of fixed objects such as buildings and street trees reflected on the road, it is also possible to remove shadows of moving vehicles and areas with uniform density such as ordinary vehicles, and the front part of each moving vehicle can be removed. Can be extracted to accurately grasp the presence state of the vehicle in the measurement area to measure the traffic flow, and can be implemented by a single-lens camera, so that the cost can be reduced. According to the second aspect of the present invention, the vehicle extracting unit included in the image processing unit further extracts an image of a portion where the two expanded images overlap each other as a plurality of continuous images at a predetermined interval, and extracts the plurality of images. The front part of the moving vehicle is extracted based on the result of the logical product of the two. For this reason, even if the shadow is divided by a trailer type truck or the like having a gap between the driver's seat and the bed portion, the shadow at the boundary can be removed. According to the third aspect of the present invention, when performing nighttime image processing, the vehicle extracting unit included in the image processing means detects a moving vehicle based on an image corresponding to a pair of headlights or a vehicle width light. Extract the front part. For this reason, by relying on the headlights and side lights of moving vehicles that are lit at night, each moving vehicle can be accurately extracted, the presence of vehicles in the measurement area can be grasped, and traffic flow can be accurately determined. Can be measured. According to the fourth aspect of the present invention, there is provided a vehicle management table for storing and managing vehicle information extracted from the measurement area of the road overhead view image for each vehicle. Then, in the vehicle tracking unit of the image processing means, each time the vehicle extraction unit extracts a new front of the moving vehicle in the measurement area, the vehicle information is registered in the vehicle management table, and the front of the moving vehicle is tracked. When the vehicle coordinates are updated, and the front of the moving vehicle disappears in the middle of the measurement area, it is determined that the vehicle is stopped and the vehicle coordinates at that time are stored.When the stopped vehicle resumes moving, the front of the moving vehicle is tracked from the stored vehicle coordinates. Restart and update vehicle coordinates. For this reason, it is possible to accurately track each moving vehicle in the measurement area while extracting the moving vehicles described above.

According to the fifth aspect of the present invention, the image processing means further includes a vehicle passage determining section. This vehicle passage determination unit provides an exit area near the vehicle exit of the measurement area, and when the vehicle tracking unit disappears after the front of the moving vehicle enters the exit area, immediately before the front of the moving vehicle enters the exit area. Performs a correlation operation between the background image of the exit area and the image of the current exit area, and deletes the vehicle information from the vehicle management table as a vehicle has passed if the calculation result exceeds a predetermined threshold, If the calculation result is equal to or less than the predetermined threshold value, it is determined that a part of the vehicle body remains, and the vehicle information is stored in the vehicle management table. For this reason, the vehicle management table accurately stores the information of the moving vehicles existing in the measurement area, and is deleted when the vehicle passes, so that the presence of the vehicle in the measurement area can be accurately determined just by looking at the vehicle management table. You can figure out. According to the invention described in claim 6, a vehicle headlight, a vehicle width light, and the like of a moving vehicle are used for vehicle extraction and vehicle tracking at night, but the vehicle passage determination unit at night determines. When a vehicle headlight or a vehicle width light passes, it disappears from the screen, and it cannot be determined whether a part of the vehicle body still remains in the exit area based on the presence or absence of the headlight or the vehicle width light. For this reason, whether or not a part of the vehicle body remains in the exit area at night is determined by checking the presence or absence of side reflections caused by a speed light, a side light, or a headlight in another lane attached to the vehicle. When the proportion of the high brightness area is large, it is assumed that a part of the vehicle body remains in the exit area. According to the invention described in claim 7, the measurement result output unit refers to the vehicle information stored in the vehicle management table,
Since at least one measurement result of the presence, the number of vehicles, or the number of passing vehicles in the measurement area is obtained and output to the outside, the above-described traffic flow measurement result is used, for example, in traffic signal control. Or provided to an in-vehicle device or the like via an optical beacon. According to the invention described in claim 8, the road overhead image from the camera installed at a high place on the road is not only an image that overlooks the front part of the moving vehicle, but also an image that overlooks the rear part of the moving vehicle. In this case, the traffic flow can be accurately measured by performing the same image processing as described above. According to the ninth aspect of the present invention, the road overhead view image from the camera installed at a high place on the road is the same image of the lane looking down on the front part of the moving vehicle and the lane looking down on the rear part of the moving vehicle. The image processing may be performed within the field of view, and the simultaneous simultaneous measurement of the traffic flow becomes possible.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of a traffic flow measuring device according to the present embodiment.

FIG. 2 is a diagram illustrating an image angle of a bird's-eye view of a vehicle traffic road from a roadside high place such as a lamppost and an area division therein.

FIG. 3 is a diagram illustrating an example of a vehicle management table.

FIG. 4 is a diagram showing an example of a captured image in which a camera overlooks a vehicle traffic route.

FIG. 5 is a diagram illustrating a forward time difference image that extracts only a portion that has changed from dark to bright between two images.

FIG. 6 is a diagram showing a backward time difference image that extracts only a portion that has changed from light to dark between two images.

FIG. 7 is a diagram showing an image obtained by expanding a forward time difference image and a backward time difference image, binarizing them with a predetermined threshold after addition, and adding the images.

8A is a diagram showing a trailer-type truck having a gap between a cab and a pallet portion, FIG. 8B is a diagram of a current image processing result, FIG. 8C is a diagram of a previous image processing result, (D) is a diagram of the result of the image processing two times before, and (e) is a diagram of the logical product of the images of (b) to (d).

FIG. 9 is a flowchart illustrating a processing operation of the traffic flow measuring device according to the present embodiment.

[Explanation of symbols]

 Reference Signs List 10 traffic flow measuring device, 12 camera, 14 image input unit, 16 image storage unit, 18 image processing unit, 20 vehicle management table, 22 measurement result output unit, 24 vehicle extraction unit, 26 vehicle tracking unit, 28 vehicle passage determination unit , 30, 32, 34, 35 lane, 36 presence measurement area, 38 exit area, 40 oncoming lane, 42, 46 moving vehicle, 44, 48 shadow.

 ──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Tomoyasu Kitagawa 1-13-8 Kamikizaki, Urawa-shi, Saitama Japan Signaling Co., Ltd. Yono Works (72) Inventor Futa Kono 1-13, Kamikizaki, Urawa-shi, Saitama No. 8 Nippon Signal Co., Ltd. Yono Office F-term (reference) HA26 5H180 BB02 CC04 CC30 DD01

Claims (9)

[Claims] 1. An image processing means for performing image processing of a bird's-eye view image from a camera installed at a high point on a road, and grasps a presence state of a vehicle in a measurement area provided in a vehicle traffic direction of a lane. A traffic flow measurement device that measures traffic flow, wherein the image processing means includes a forward time difference image that extracts only a portion that changes from dark to bright between two road overhead images that are continuous at predetermined time intervals; A reverse direction time difference image for extracting only a portion changed from light to dark, and a vehicle extracting unit for extracting a front portion of the moving vehicle based on an image of a portion where the two expanded images overlap each other; Traffic flow measuring device. 2. The vehicle extracting section further extracts an image of a portion where the two dilated images overlap each other as a plurality of continuous images at a predetermined interval, and obtains a logical product of the plurality of images. The traffic flow measuring device according to claim 1, wherein a front portion of the moving vehicle is extracted. 3. The vehicle extraction unit according to claim 1, wherein when performing nighttime image processing, the vehicle extraction unit extracts a front portion of the moving vehicle based on an image corresponding to a pair of headlights or a vehicle width light. 2. The traffic flow measuring device according to 1. 4. A vehicle management table for storing and managing, for each vehicle, vehicle information extracted from within a measurement area of the road overhead view image, wherein the image processing means includes: Each time a new moving vehicle front part is extracted, the vehicle information is registered in the vehicle management table, and the vehicle coordinates are updated while tracking the moving vehicle front part. When the vehicle disappears, the vehicle is determined to be a stopped vehicle, the vehicle coordinates at that time are stored, and when the stopped vehicle restarts moving, the vehicle tracking unit that resumes tracking of the front of the moving vehicle from the stored vehicle coordinates and updates the vehicle coordinates is further provided. The traffic flow measuring device according to any one of claims 1 to 3, wherein 5. An image processing means for providing an exit area near a vehicle exit in the measurement area, wherein when the vehicle tracking section disappears after the front section of the moving vehicle enters the exit area, the front section of the moving vehicle exits the exit area. A correlation calculation is performed between the background image of the exit area immediately before entering the area and the image of the current exit area, and if the calculation result exceeds a predetermined threshold value, the vehicle information is determined to be passed by the vehicle. A vehicle passage determination unit that deletes the vehicle information from the vehicle management table and stores the vehicle information in the vehicle management table assuming that a part of the vehicle body remains if the calculation result is equal to or less than a predetermined threshold value. The traffic flow measuring device according to claim 4, characterized in that: 6. An image processing means for providing an exit area near a vehicle exit of the measurement area, extracting a vehicle at night by the vehicle extracting section according to claim 3, and a vehicle front section by the vehicle tracking section. If the vehicle disappears after entering the exit area, the exit area is binarized by brightness, and if the ratio of the high brightness area in the exit area is equal to or less than a predetermined threshold, the vehicle information is determined to be passed by the vehicle. The vehicle management table further includes a vehicle passage determination unit that deletes the vehicle information from the vehicle management table and stores the vehicle information in the vehicle management table as a part of the vehicle body remains if a predetermined threshold is exceeded. The traffic flow measuring device according to claim 4, wherein 7. Based on the vehicle information stored in the vehicle management table, at least one of the presence of a vehicle, the number of vehicles, and the number of vehicles passing in a measurement area.
7. The apparatus according to claim 4, further comprising a measurement result output unit for obtaining one measurement result and outputting the measurement result to the outside.
The traffic flow measuring device according to any one of the above. 8. An image of a bird's-eye view of a road from a camera installed at a high point on the road is processed using a bird's-eye view image of a rear part of the moving vehicle instead of a bird's-eye view image of a front part of the moving vehicle. The traffic flow measuring device according to any one of claims 1 to 7. 9. A bird's-eye view image from a camera installed at a high point on the road includes a lane that looks down on the front part of the moving vehicle and a lane that looks down on the rear part of the moving vehicle in the same image field of view. The traffic flow measurement device according to any one of claims 1 to 7, wherein the process allows the simultaneous measurement of the opposite direction.