JP2005353004A - Vehicle traveling measurement system and vehicle tracking method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable assignment of a common vehicle ID to the same vehicle among a plurality of cameras in a vehicle traveling measuring system including the plurality of cameras set on a road to mutually continue the monitoring sections of adjacent cameras to detect a vehicle traveling in the monitoring section of each camera. <P>SOLUTION: The plurality of cameras are set on the road R so that monitoring sections S of adjacent cameras are mutually continued, and images of the respective monitoring sections S are taken by the plurality of cameras. Based on the taken image signals of the cameras, the position (x, y) of a vehicle traveling on the road is detected at predetermined intervals to collect vehicle data containing the position (x, y) of the vehicle. When the vehicle positions (x, y) detected by the adjacent cameras and the detection times of the vehicle positions in an overlapping monitoring section W are close to each other, the data of the vehicle captured by the image processing part of the downstream-side camera and the data of the vehicle captured by the image processing part of the upstream-side camera are accumulated as data of the same vehicle. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a vehicle travel measurement system that installs a plurality of cameras on a road and detects a vehicle traveling in a monitoring section of the camera, and a vehicle tracking method that tracks a vehicle traveling in a plurality of monitoring sections. .

In order to detect a vehicle traveling on a road, conventionally, a technique of installing an ultrasonic type or loop coil type detector on the road has been the mainstream. In these methods, a method of detecting the passage of a vehicle in a spot manner and grasping a traffic flow state based on a change in the passing traffic volume per unit time has been used.
However, in recent years, a technique for detecting a vehicle by installing a camera on a road and processing an image obtained from the camera has been developed. Since it is possible to measure a wide range within the photographing range of the camera, it has become possible to measure the behavior of individual vehicles in addition to the above-described spot-like measurement.
JP 2002-329195 A

However, in a conventional vehicle travel measurement system using a camera, it is difficult to identify a vehicle between adjacent cameras because image processing is performed for each camera.
Therefore, the present invention provides a vehicle travel measurement system in which a plurality of cameras are installed on a road so that adjacent camera monitoring sections are continuous, and a vehicle traveling in the monitoring section of each camera is detected. An object of the present invention is to provide a vehicle that can give a common vehicle ID to the same vehicle.

  Another object of the present invention is to provide a vehicle tracking method in the system.

  The vehicle travel measurement system of the present invention is connected to a plurality of cameras installed on the road, and detects vehicle data including the position of the vehicle traveling on the road at predetermined time intervals based on video signals of the cameras. An image processing unit and a vehicle data integration / storage unit that collects and accumulates vehicle data from each of the image processing units, and the vehicle data integration / storage unit is an overlapping monitoring section of adjacent cameras, When the vehicle position included in the vehicle data collected from the image processing unit of the adjacent camera and the detection time of the vehicle position are close, the vehicle data and upstream side of the vehicle captured by the image processing unit of the downstream camera The vehicle data captured by the image processing unit of the camera is accumulated as vehicle data of the same vehicle (claim 1).

With this configuration, the vehicle data integration / accumulation unit can integrate the vehicle data detected independently by each image processing unit based on the approximation of the vehicle position and the detection time of the vehicle position. Therefore, the vehicle can be identified and tracked over a plurality of continuous monitoring sections.
The image processing unit may output the vehicle data of the plurality of vehicles as data of the same vehicle when the vehicle positions and the detection times of the vehicle positions of the plurality of vehicles respectively detected in the monitoring section are close. Preferred (claim 2). This process can be integrated based on the approximation of the vehicle position and the detection time of the vehicle position when the vehicle data is erroneously detected or double-detected within the monitoring section of one camera. This integration process further improves the accuracy of the vehicle data and facilitates vehicle identification over a plurality of continuous monitoring sections.

  The image processing unit calculates an estimated speed of the vehicle when the vehicle data is missing at any detection time in the monitoring section, and interpolates the position data of the vehicle based on the calculated estimated speed (Claim 3). In this process, when vehicle data is missing at a certain time within a monitoring section of one camera, an estimated speed of the vehicle is calculated based on other data of the vehicle, and based on the calculated estimated speed, This is a process of interpolating the position data of the vehicle. This interpolation further improves the accuracy of the vehicle data and facilitates vehicle identification over a plurality of consecutive monitoring sections.

  Moreover, it is preferable to exclude such vehicle data because performing each of the above processes based on erroneously detected vehicle data leads to erroneous results. Therefore, the image processing unit excludes vehicle data of a vehicle with a small number of vehicle detections from the output target (Claim 4), or excludes vehicle data having a large variation in detected vehicle position from the output target. (Claim 5).

  The vehicle travel measurement system of the present invention is connected to a camera installed on the road, and detects vehicle data including the position of the vehicle traveling on the road at predetermined intervals based on the video signal of the camera. An image processing unit, and when the vehicle position and the detection time of the vehicle position of a plurality of detected vehicles are close in the monitoring section, the image processing unit stores the vehicle data of the plurality of vehicles in the same vehicle (Claim 6). This vehicle travel measurement system can be integrated based on the approximation of the vehicle position and the detection time of the vehicle position when vehicle data is erroneously detected or double-detected within the monitoring section of one camera. . Therefore, it is possible to prevent data of one vehicle from being erroneously output as data of a plurality of vehicles, and to obtain highly accurate vehicle data.

  The vehicle travel measurement system of the present invention is connected to a camera installed on the road, and detects vehicle data including the position of the vehicle traveling on the road at predetermined time intervals based on the video signal of the camera. A processing unit, and the image processing unit calculates an estimated speed of the vehicle when vehicle data is missing at any detection time in the monitoring section, and based on the calculated estimated speed, The vehicle position data is interpolated (claim 7). This vehicle travel measurement system calculates an estimated speed of the vehicle by other data of the vehicle when vehicle data is missing within the monitoring section of one camera, and based on the calculated estimated speed, Interpolate the position data of the vehicle. Therefore, it is possible to output continuous vehicle data with no omission within the monitoring section.

  The vehicle tracking method of the present invention is a method invention according to the same invention as that of the vehicle travel measurement system according to claim 1 (claim 8).

  As described above, according to the present invention, the vehicle can be identified and tracked over a plurality of continuous monitoring sections, and there is a specific effect that more accurate measurement of vehicle travel can be performed.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
FIG. 1 is a block configuration diagram of the vehicle travel measurement system 1. The vehicle travel measurement system 1 includes a plurality of CCTV (Closed-Circuit Television) cameras (hereinafter referred to as “cameras”) 2, an image processing unit 3, an interface unit 4, and a data integration / storage unit 5 that monitor a road. .

Each camera 2 and the image processing unit 3 are connected by a cable.
The image processing unit 3 detects the position of the vehicle traveling on the road R based on the video input signal from each camera 2, and performs processing such as speed calculation and data interpolation based on the detected position of the vehicle. The interface unit 4 collects vehicle data (data including vehicle ID, position, speed, etc. is referred to as “vehicle data”, hereinafter the same) output from each image processing unit 3. The data integration / storage unit 5 integrates and stores vehicle data for each aggregated image processing unit 3.

  All or part of the vehicle detection and tracking processing functions in the image processing unit 3 are realized by the computer of the image processing unit 3 executing a program recorded in a predetermined medium such as a ROM, an ASIC, or a hard disk. The data integration function in the data integration / accumulation unit 5 is similarly performed by the computer of the data integration / accumulation unit 5.

  FIG. 2 is a diagram showing the camera 2 installed on the highway R. As shown in FIG. Each camera 2 is installed on a pole or the like along the road R, and monitors a section on the road R (referred to as “monitoring section S”, to which a subscript is appropriately assigned). It is desirable that adjacent monitoring sections S overlap each other, and the overlapping area is referred to as “overlapping section W” (appropriate subscripts are given as appropriate). The overlapping section W belongs to any adjacent monitoring section S.

  FIG. 3 is a plan view of the highway R where the camera 2 is installed. The direction crossing the road R is defined as x direction, and the direction along the road R is defined as y direction. The origin of the x coordinate is the white line between the traveling lane and the road shoulder. The y direction is taken along the curve if the road R is curved. The origin of the y coordinate is the measurement start point of the camera 2 installed on the uppermost stream, and the y coordinate is expressed as a road from that point.

As an example of the embodiment, the xy coordinates may be determined based on the direction regardless of the road direction or the presence or absence of a curve.
The screen actually captured by the camera 2 does not correspond to the xy plane of the road R. This is because the camera 2 looks down the road R from a high position, as can be seen from FIG. Therefore, the conversion formula between the coordinates of the screen captured by the camera 2 and the xy coordinates of the actual road R is incorporated into the computer of the image processing unit 3 using the height of the camera 2 and the angle at which the camera 2 is looked down as parameters. (See JP 2003-042760, etc.).

Time synchronization between the image processing units 3 will be described with reference to FIGS. 4 and 5.
Each image processing unit 3 uses an independent synchronization signal in principle, but in order to perform data integration processing in the data integration / accumulation unit 5, the time of vehicle data acquired by each image processing unit 3 is synchronized. There is a need. Therefore, as shown in FIG. 4, the data integration / accumulation unit 5 periodically transmits a time signal to the interface unit 4. The data integration / accumulation unit 5 is preferably connected to a radio clock or an external time server so as to maintain an accurate time. As shown in FIG. 5, the interface unit 4 transmits a time signal from the image processing unit 3 to each image processing unit 3 when an event occurs such as when the image processing unit 3 is activated or when a predetermined time elapses. As a result, time synchronization between videos is achieved.

  However, since frequent synchronization processing increases the load on each image processing unit 3, the "certain time" is, for example, about once per hour. Thus, the image processing units 3 do not always synchronize between the cameras 2, and the time of the vehicle data of the same vehicle does not always coincide completely between adjacent images. In the case of NTSC standard video, the video sent from the camera 2 is input to the image processing unit 3 at a cycle of about 30 frames per second. Since the frame synchronization signal is independent for each camera 2, each image processing unit 3 cannot perfectly match the timing of image processing. Accordingly, the data integration / accumulation unit 5 requires a certain amount of time to determine that the vehicle data of different cameras are connected at the same time (details will be described later).

FIG. 6 is a general outline flowchart for collectively explaining vehicle data generation and data interpolation processing performed by the image processing unit 3 and vehicle data integration processing performed by the data integration / accumulation unit 5.
First, the image processing unit 3 detects each vehicle in the monitoring section S (step S0). This process is a conventionally known process, but an example will be briefly described.

  The image input interval is, for example, every 0.2 seconds. A difference between the image input from the camera 2 and the background image is obtained to obtain an image showing only moving objects such as a vehicle. A binarization process is performed on this image with an intermediate luminance as a threshold value, thereby detecting the size of an object shown in the image. Is determined. The vehicle tail of the object determined to be this vehicle is tracked.

Here, the end of the vehicle is the center position of the rear surface of the vehicle as shown in FIG. The y coordinate at the end of the vehicle can be determined from the luminance difference between the object determined to be a vehicle and the road surface. The x coordinate at the end of the vehicle is calculated based on the x coordinate of the center of gravity of the object determined to be a vehicle.
In order to track the detected vehicle, the image processing unit 3 registers vehicle data. Table 1 is a table showing main registration items of vehicle data in each image processing unit 3.

As described above, the image processing unit 3 is premised on tracking the end of the vehicle extracted in the monitoring section S, and therefore, when the end of the vehicle cannot be measured (there is no sufficient inter-vehicle distance; For example, in the case of hiding the vehicle data, the vehicle data is not output.
Next, the image processing unit 3 calculates the speed and acceleration in the y direction for each vehicle (step S1). This speed calculation is performed by dividing the difference in the y coordinate of the vehicle between two times by the time. The acceleration is calculated from the difference in vehicle speed between two times. By the way, since speed and acceleration include measurement errors, it is desirable to smooth them using a statistical method. As a smoothing method, the moving average method that averages the values detected over the past several times, or the exponential smoothing method that calculates the current exponential smoothing value by weighted averaging of the previous exponential smoothing value and the current measured value Etc. can be adopted. The number of times to take an average in the moving average method is considered to be about the number of times corresponding to 1 second (for example, 5 times if one processing period is 0.2 seconds). It is desirable to decide according to utilization experience. It is desirable that the weighting coefficient in the exponential smoothing method is also determined according to the measurement resolution determined by the angle-of-view condition and the data utilization experience.

  Next, the velocity and acceleration in the x direction are calculated for each vehicle (step S2). The calculation method is the same as described above. The detection of the x coordinate at the end of the vehicle is calculated based on the center of gravity of the area determined to be a vehicle as described above, and is relative to the y coordinate at the end of the vehicle obtained relatively stably from the luminance difference between the vehicle and the road surface. The variation is large. Therefore, in an area or traffic situation in which lane changes are hardly seen, the speed in the x direction is regarded as 0, and instead, an average value of x coordinates is calculated to identify the lane.

Next, excluded data is excluded (step S3). Data that seems to be erroneously detected or data that has a small number of tracking times and that is judged to have poor data interpolation accuracy until the next overlapping section W are excluded.
There are the following methods as specific methods for determining “non-target data”. (1) Exclude those with a small number of vehicle detections. When the front and rear traveling vehicles and vehicles traveling in the adjacent lane appear to overlap each other, the vehicle ahead is not always detected, and the number of vehicle detections is expected to decrease. In addition, in the case where a non-vehicle end is double detected as the end of the vehicle, it is unlikely that the same part of the vehicle will continue to be erroneously detected, so this is also expected to reduce the number of vehicle detections. Therefore, the vehicle data is excluded when the number of times of vehicle detection is less than or equal to a predetermined ratio compared to the number of times of full detection.

  Table 2 is a table showing a result of detecting a vehicle with two adjacent cameras 2 upstream and downstream (assuming that a vehicle that has passed through the monitoring section S is accurately known by visual observation of an image).

  Of the vehicles that have been detected correctly, the number of tracking that is as small as 3 or less and 5 or less is small. For example, in the downstream camera 2, out of 83 vehicles that are correctly detected, 9 vehicles are tracked less than 5 times. For those that have been mistakenly detected as vehicles, the number of tracking is relatively small when the number of tracking is as small as 3 or less and 5 or less. For example, in the upstream camera 2, out of 24 cases in which the shadow is detected as a vehicle, 21 cases have a tracking count of 5 times or less.

  Further, (2) there is a method of determining an excluded vehicle from the detected variation in the vehicle position. For a plurality of vehicles that have passed through the monitoring section S, those that have been detected correctly and those that have been erroneously detected except for the end of the vehicle are collected. The difference in speed in the y direction was determined. F test was performed with the equal variance of both groups as the null hypothesis. As a result, as shown in Table 3, both the velocity in the x direction and the acceleration in the y direction are rejected at the significance level of 1%, and the dispersibility between the positive detection data and the misdetected data other than the end of the vehicle is significant. It is shown that there is a difference.

Next, vehicle data interpolation processing is performed for each camera in the monitoring section S (step S4). This process is performed in each image processing unit 3 and is a process of interpolating the position and speed of the vehicle at that time when there is a lack in vehicle detection. The data interpolated in this way with respect to the actually detected vehicle data is referred to as “virtual tracking data”.
Hereinafter, the vehicle data interpolation process in the monitoring section S will be described with reference to a detailed flowchart (FIG. 7). First, vehicle data registered in the monitoring section S (see Table 1) is searched, and those vehicles are taken out one by one (step T1).

  It is determined whether or not the vehicle is detected in the current processing period (step T2). If detected in the current processing period, if there is a missing data section so far, the trajectory data in the missing section is estimated from the difference between the previous trajectory data and the current data, and registered as virtual tracking data ( Step T7). If not detected in the current processing period, the following data interpolation processing is performed.

First, it is examined whether or not the detection has been interrupted more than the specified number of times (step T3). If the detection has been interrupted more than the specified number of times, it is checked whether or not the trajectory before the interruption has reached the handover section to the next camera 2 (step T4). If not, the estimated speeds in the x and y directions are calculated based on the trajectory data so far (step T5).
As described in steps S1 and S2, this speed estimation method is performed based on the position detected in the past processing period in which target data existed. As described above, it is desirable to smooth this speed by using a statistical method. For example, if the n-time moving average method is adopted, n processing periods in which target data existed in the past are selected, and the average of those data is taken.

  The virtual tracking data is created by extending the vehicle position to the hand over section to the next camera 2 (step T6). This creation method assumes that the vehicle travels at the estimated speed even in a processing period in which the target data no longer exists, and predicts the position of the vehicle in each processing period. The position is plotted for each time. And the position of the said vehicle is registered into vehicle data. Since the position of this vehicle is based on the result of virtual tracking, a virtual flag is set (see Table 1).

  Note that when there is a road exit in the monitoring section S, the vehicle may get off. Even in this case, the tracking of the vehicle is interpolated to obtain virtual tracking data. The reason is that it is difficult to distinguish whether the vehicle does not exist or whether the vehicle cannot be tracked due to a false detection. This is because the virtual data when there is no vehicle can be deleted by the vehicle connection determination in the overlapping section W, which will be described later.

  FIG. 8 is a graph for explaining a speed estimation method using this statistical method. The horizontal axis represents time t, and the vertical axis represents y-direction coordinates. The scale on the horizontal axis represents the processing period. Black circles indicate data that could be detected in the past. White circles represent data that could not be detected. Since it cannot be detected at the current time and the time before one processing period, it is a white circle. For example, if the three-time moving average method is adopted, the moving average is calculated based on the data of the three black circles that have been detected most recently, and the moving average curve is extended in a straight line to obtain the white circle data. Is estimated.

The above data interpolation process is performed for each registered vehicle (step T8). In this way, the interpolation of the vehicle data in each camera in the monitoring section S is completed.
If there is a vehicle that has not been registered so far, the vehicle is newly registered (step T9).
Next, the same vehicle determination process in each camera in the monitoring section S is performed (step S5).

  The vehicle data interpolation process described above is a process of interpolating the missing vehicle data based on past vehicle data when the vehicle data is missing. However, vehicle data of the same vehicle may be detected redundantly due to erroneous detection. For example, this is a case where a shadow of a vehicle is erroneously detected as a vehicle. In such a case, the process for determining the same vehicle and removing redundant data is the same vehicle determination process described below.

In FIG. 9, the detailed flowchart for demonstrating the same vehicle determination process in the monitoring area S is shown. First, vehicle data registered in the monitoring section S is searched, and all those vehicles are taken out (step U1). At this time, virtual tracking data is also included.
Next, it is determined whether another vehicle exists at the same time and at the same position (step U2). The determination method is performed by the logical product of two elements. Here, the “same position” means that the distance within a certain error range is regarded as the same position, not the exact same position. This is because there is an error in the vehicle detection position by the camera 2. For example, the y-coordinate is ± 5 m, the x-coordinate is ± 2 m, and so on.

  If another vehicle exists at the same time and the same position, these vehicles are determined to be the same vehicle, and the same vehicle ID is assigned again (step U3). When multiple vehicles are at the same position, the closer one is tied. Then, the overlapping vehicle data is deleted. If either one is virtual data, the virtual data is deleted, and if both are real data, the remote data is deleted (step U4).

In addition to the determination of the same position, “same speed” may be added as an AND determination element. For example, if the speed in the y direction is within ± 20 km / h, it is regarded as the same speed.
When the vehicle ID is reassigned, various rules for determining the vehicle ID can be considered. For example, a vehicle that has been tracked for a longer time may be regarded as a real vehicle, and the vehicle ID may be assigned. Further, if one of the vehicle position data existing at the same time and at the same position is data with a virtual flag, the vehicle ID related to the virtual data may be deleted, and the vehicle ID of a non-virtual real vehicle may be employed.

FIG. 10 is a graph showing a state in which a plurality of vehicles are present at the same time and at the same position. The horizontal axis represents time t and the vertical axis represents y-coordinate. When the vehicle position is within a certain section at the same time, it is regarded as the same vehicle. In the graph, when the vehicle of ID _{1 and} the vehicle of ID ₂ are regarded as the same vehicle, they are unified with the vehicle of ID ₁ detected from before. When the vehicle of ID _{3 and} the vehicle of ID ₄ are regarded as the same vehicle, ID ₃ is virtual tracking data, so it is unified with the vehicle of ID ₄ .

Next, the vehicle connection determination process in the overlap section W between adjacent cameras will be described (step S6).
In one monitoring section S, as described above, independent vehicle tracking is performed for each monitoring section S. Accordingly, different vehicle IDs are usually given to the same vehicle in the overlapping section W. The vehicle position is also discontinuous data in the overlapping section W. Therefore, this process takes over the vehicle data such as the vehicle ID of the same vehicle from one monitoring section S to the next monitoring section S.

FIG. 11 shows a detailed flowchart for explaining the vehicle connection determination process in the overlapping section W.
As shown in FIGS. 4 and 5, the data integration / accumulation unit 5 regularly receives each vehicle data by all the image processing units 3 (step V <b> 1).
Next, based on these vehicle data, it is determined whether or not a plurality of vehicles are present at the same time and at the same position among the vehicles detected by the respective image processing units 3 (step V2). Here, “the same time” means that the time within a certain error range is regarded as the same time, not the exact same time. This is because, as described above, time synchronization is not always performed between the image processing units 3. For example, up to ± 0.2 seconds is regarded as the same time. Therefore, if the detection processing time is every 0.2 seconds, even if there is a difference up to one processing time before and after, the same time is considered. The meaning of “same position” is as described above in the same vehicle determination process. Further, the meaning of “plurality” means that each image processing unit 3 has detected. The plurality of vehicles detected by the same image processing unit 3 are performed in the same vehicle determination process in the monitoring section S as described in the detailed flowchart of FIG.

  If there are a plurality of vehicles at the same time and at the same position, these vehicles are determined to be the same vehicle, and the same vehicle ID is assigned again (step V3). The ID to be given is preferably a vehicle ID on the upstream side. Thereby, the travel history on the upstream side of the vehicle can be taken over as it is. If the data is duplicated, one of the data is deleted (step V4).

Further, the position of the vehicle in the overlapping section W is unified with the position detected by the image processing unit 3 corresponding to the camera 2 on the downstream side instead of the upstream side.
FIG. 12 is a graph showing a state in which a plurality of vehicles exist at the same time and at the same position in the overlapping section W ₁₂ of the monitoring sections S ₁ and S ₂ . The horizontal axis represents time t and the vertical axis represents y-coordinate. If the vehicle position is within the dashed circle, it is considered the same vehicle. When the ID ₁ vehicle detected in the upstream monitoring section S ₁ and the ID ₂ vehicle detected in the downstream monitoring section S ₂ are regarded as the same vehicle, the ID detected in the monitoring section S ₁ Unified to ₁ . If the ID ₃ vehicle detected in the upstream monitoring section S ₁ and the ID ₄ vehicle detected in the downstream monitoring section S ₂ are regarded as the same vehicle, the ID ₃ detected in the monitoring section S ₁ It is unified.

In this way, data of the same vehicle detected in the adjacent monitoring section S can be connected and integrated.
Although the embodiments of the present invention have been described above, the embodiments of the present invention are not limited to the above-described embodiments. For example, as a criterion for “same time, same position”, in addition to a criterion that two times are closer than a threshold time and two positions are closer than a threshold distance, in a multidimensional space The concept of “distance” may be introduced. For example, the time difference of the data is Δt, the distance difference of the x coordinate is Δx, the distance difference of the y coordinate is Δy, and weighting coefficients k, m, n are attached to these,
k | Δt | + m | Δx | + n | Δy |
May be determined to be the same data when the score is obtained and the score falls within a certain score. Of course, “same speed” may be added to make it four-dimensional. If there are a plurality of identical vehicle candidates, the vehicle with the smallest score may be determined as a vehicle. In addition, various modifications can be made within the scope of the present invention.

1 is a block configuration diagram of a vehicle travel measurement system 1. FIG. It is a figure which shows the camera 2 installed on the highway R, and its monitoring section S. FIG. It is a top view of the highway R where the camera 2 was installed. It is a normal synchronous process sequence diagram. It is a synchronous processing sequence when a specific event occurs. 4 is an overall outline flowchart for collectively explaining processing such as vehicle data generation and data interpolation performed by the image processing unit 3 and vehicle data integration processing performed by the data integration / accumulation unit 5; 4 is a detailed flowchart showing a vehicle data interpolation process in a monitoring section S. It is a graph for demonstrating the speed estimation method using a statistical method. 4 is a detailed flowchart showing identical vehicle determination processing in a monitoring section S. 4 is a graph showing a state in which a plurality of vehicles are present at the same position in a monitoring section S. 5 is a detailed flowchart showing a vehicle connection determination process in an overlapping section W. It is a graph which shows a mode that several vehicles exist in the same time in the overlap area W, and the same position.

Explanation of symbols

1 Vehicle Travel Measurement System 2 Camera 3 Image Processing Unit 4 Interface Unit 5 Data Integration / Storage Unit

Claims (8)

In the vehicle travel measurement system for installing a plurality of cameras on the road so that the monitoring sections of adjacent cameras are continuous, and detecting a vehicle traveling in the monitoring section of each camera,
A plurality of image processing units connected to the plurality of cameras, respectively, for detecting vehicle data including a position of a vehicle traveling on a road at predetermined time intervals based on video signals of the cameras;
A vehicle data integration / accumulation unit that collects and accumulates vehicle data from each of the image processing units;
The vehicle data integration / accumulation unit is a monitoring section where adjacent cameras overlap, and the vehicle position included in the vehicle data respectively collected from the image processing unit of the adjacent camera and the detection time of the vehicle position are close The vehicle data of the vehicle captured by the image processing unit of the downstream camera and the data of the vehicle captured by the image processing unit of the upstream camera are accumulated as vehicle data of the same vehicle. Vehicle running measurement system.   The image processing unit outputs the vehicle data of the plurality of vehicles as data of the same vehicle when the vehicle positions and the detection times of the vehicle positions of the plurality of vehicles detected in the monitoring section are close. The vehicle travel measurement system according to claim 1.   The image processing unit calculates an estimated speed of the vehicle when the vehicle data is missing at any detection time in the monitoring section, and interpolates the position data of the vehicle based on the calculated estimated speed The vehicle travel measurement system according to claim 1, wherein   The vehicle travel measurement system according to any one of claims 1 to 3, wherein the image processing unit excludes vehicle data of a vehicle with a small number of vehicle detections.   4. The vehicle travel measurement system according to claim 1, wherein the image processing unit determines an excluded vehicle based on the detected variation in the vehicle position. 5. In a vehicle travel measurement system that installs a camera on a road and detects a vehicle traveling in a monitoring section,
An image processing unit that is connected to the camera and detects vehicle data including a position of a vehicle traveling on a road at predetermined time intervals based on a video signal of the camera;
The image processing unit outputs the vehicle data of the plurality of vehicles as the data of the same vehicle when the vehicle positions and the detection times of the vehicle positions of the plurality of vehicles detected in the monitoring section are close. A vehicle running measurement system characterized by being. In a vehicle travel measurement system that installs a camera on a road and detects a vehicle traveling in a monitoring section,
An image processing unit that is connected to the camera and detects vehicle data including a position of a vehicle traveling on a road at predetermined time intervals based on a video signal of the camera;
The image processing unit calculates an estimated speed of the vehicle when the vehicle data is missing at any detection time in the monitoring section, and interpolates the position data of the vehicle based on the calculated estimated speed A vehicle travel measurement system characterized by the above. Install multiple cameras on the road so that the monitoring sections of adjacent cameras are continuous,
Each surveillance section is photographed by the plurality of cameras,
Based on the captured video signal of the camera, the position of the vehicle traveling on the road is detected every predetermined time,
In the vehicle tracking method for collecting and storing vehicle data including the position of the vehicle,
In the overlapping monitoring section, when the vehicle position detected by the adjacent camera and the detection time of the vehicle position are close, the data of the vehicle captured by the image processing unit of the downstream camera and the image processing of the upstream camera The vehicle tracking method characterized by accumulating the data of the said vehicle caught by the part as data of the same vehicle.

Images