JP4858761B2 - Collision risk determination system and warning system - Google Patents

Description

  The present invention relates to a collision risk determination system and a warning system that acquire image data by photographing and determine a collision risk based on the acquired image data.

In the conventional sensor technology that monitors road traffic and measures traffic flow, macro observations that measure the traffic volume by measuring the number of passing vehicles with a vehicle detector were the mainstream. The purpose of this macro observation technique is to detect a vehicle and output the traffic volume and average speed of a certain time section, and it cannot be used to analyze traffic accidents.
On the other hand, recent developments in image processing technology have made it possible to directly grasp the behavior of individual vehicles. If this image processing technology is used, a large amount of data can be processed, and individual characteristics do not affect the analysis result. Therefore, it is possible to quantitatively analyze the behavior of a large number of vehicles.
JP-A-5-307695 Iida, Uno, Itsubo, Kakinuma "Modeling of conflict analysis and lane change in the weaving part", Proc. 24th Civil Engineering Research Proceedings, pp. 305-306, November 2001

In order to reduce traffic accidents, it is necessary to know whether or not a vehicle is traveling safely. To that end, it is necessary to calculate the relationship between the vehicle and neighboring vehicles.
Therefore, the present invention uses an image processing technique to grasp the relationship between a vehicle of interest and a nearby vehicle, thereby supporting the safe driving of the vehicle and preventing the occurrence of a traffic accident. And to provide a warning system.

The collision risk determination system of the present invention includes a camera for acquiring image data by photographing, and a measuring device for determining a collision risk based on the acquired image data. Vehicle recognition means for recognizing a vehicle traveling on the road based on the above, and when a plurality of vehicles are recognized, the speed and inter-vehicle distance of each vehicle are calculated, and the collision is performed based on the calculated speed and inter-vehicle distance. calculating means for calculating a risk index, possess and output means for outputting a value of the calculated collision risk index, said calculating means, if a plurality of vehicles at the camera shot from behind obliquely, the The position 1 on the screen closest to the camera of the vehicle traveling ahead of the plurality of vehicles is measured, and the position 2 on the screen farthest from the camera of the vehicle following the vehicle traveling ahead is measured. And Based on the coordinates of the position 2, coordinates on the road obtained by obliquely projecting the portion corresponding to the position 2 of the vehicle to the ground are calculated, and a predetermined distance is subtracted from the coordinates on the road, thereby obtaining the position 2 And the inter-vehicle distance is calculated from the position 1 and the corrected position 2 .

Here, the “collision risk index” refers to an index that objectively represents the degree of risk that a plurality of vehicles are likely to cause a collision accident. It is also called a traffic conflict evaluation index.
According to this collision risk determination system, a collision risk index between a plurality of vehicles can be automatically calculated based on the speed and the inter-vehicle distance. Based on this collision risk index, it is possible to determine the risk of collision of a plurality of vehicles. By outputting the value of the collision risk index, it is possible to support safe driving of the vehicle and prevent a traffic accident from occurring.

The inter-vehicle distance is to measure the position 1 of the side (Kuzumo) close to the preceding vehicle camera, a position 2 on the far side (headway) to the following vehicle camera is measured, so determined by the difference between these two positions Then , based on the fact that the camera is photographing the vehicle at an angle, there may be an error that the position on the side farther from the camera of the following vehicle is evaluated farther than it actually is. Therefore, if the position 2 is corrected using the predetermined distance and the inter-vehicle distance is calculated from the position 1 and the corrected position 2, the collision risk index can be accurately calculated.

The collision risk index includes an inter-vehicle distance, a travel distance from when the preceding vehicle starts to decelerate until it stops, an idle travel distance until the succeeding vehicle notices the deceleration of the preceding vehicle, and the succeeding vehicle starts decelerating. It can be calculated based on the travel distance from the time until the vehicle stops.
If the system further includes a storage means for storing the collision risk index calculated by the calculation means, a large amount of collision risk index data can be created, which is useful for analyzing traffic jams and traffic accidents. For example, it is possible to determine the necessity of installing a road sign from the viewpoint of traffic engineering based on the accumulated data of the collision risk index. It is also possible to determine the necessity of traffic control.

  The collision risk warning system according to the present invention includes the collision risk determination system and a transmission unit that transmits warning information to a road administrator when a collision risk is determined based on a collision risk index. It is. With this system, a vehicle with a high risk of collision can be detected in real time, so the vehicle can be watched in advance in conjunction with the monitoring monitor of the road manager. Action can be taken quickly.

In addition, when a system including a transmission unit that transmits warning information to a vehicle when a collision risk is determined based on a collision risk index, a driver for a vehicle with a high risk of collision and its subsequent vehicles is provided. Information can be provided immediately before you notice.
If the time when the warning information is transmitted to the vehicle is later than the time when the preceding vehicle passes the installation position of the transmission means, it is not necessary to give troublesome information to the preceding vehicle and the preceding vehicle. Smooth traffic.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
FIG. 1 shows an installation diagram of the collision risk judgment system of the present invention.
The camera 2 is installed at a predetermined height so as to overlook the road surface above the road. The field of view of the camera 2 is on one lane of the road, its width is substantially equal to the lane width, and the length D along the road is, for example, about 100 m.

The captured image of the camera 2 is input to the measuring device 4 through the video cable 3. The measuring device 4 has a communication means, and can provide a measurement result to a traffic light, a display board, or a remote traffic control center.
FIG. 2 is a block diagram illustrating a configuration example of the measuring device 4.
The measuring device 4 captures the image of the camera 2, determines the presence or absence of the vehicle in the vehicle detection measurement window set on the screen of the camera 2, and when the vehicle is detected, calculates the vehicle position coordinates, Executes calculation processing of collision risk index during deceleration. The camera 2 and the measuring device 4 may be integrated.

As shown in FIG. 2, the measurement device 4 includes an image processing unit 41 and an event determination unit 42.
The image processing unit 41 includes an input unit that provides an interface for capturing an image of the camera 2, an A / D conversion unit that digitally converts an image signal, an image memory that temporarily stores the digitally converted image data, and the like. ing. Then, processing such as vehicle discrimination and vehicle position coordinate calculation is performed based on the image data.

The event determination unit 42 calculates the vehicle speed and the inter-vehicle distance based on the calculated vehicle position coordinates and the like, and obtains a collision risk index during sudden deceleration.
The determination result of the event determination unit 42 is transmitted to a traffic light, a display board, a traffic control center, etc. through a communication line.
FIG. 3 is a flowchart illustrating an image processing method performed by the image processing unit 41. All or some of the functions of the respective units of the image processing unit 41 are realized by the computer of the measuring device 4 executing a program recorded on a predetermined medium such as a CD-ROM or a hard disk.

The image processing unit 41 initializes various variables for projective transformation from the coordinate system on the screen to the coordinate system in the real space on the road (step S1).
An example of the projective transformation method will be described below with reference to FIG. This conversion method is described in detail, for example, in JP-A-2003-42760.
A coordinate system on the road surface (referred to as a local coordinate system) is defined by the X, Y, and Z axes, and the coordinates of the camera installation position O (projection center) are (X0, Y0, Z0). The Y axis is assumed to be along the traveling direction of the road. The X axis is parallel to the road surface and is perpendicular to the road traveling direction. The Z axis is perpendicular to the road surface.

The camera coordinate system rotates counterclockwise by the angle of κ with respect to the positive direction of the Z axis, similarly rotates by the angle of φ around the Y axis, and tilts by rotating the angle of ω around the X axis. Suppose that The coordinate system of the tilted camera is defined by the x-axis, y-axis, and z-axis.
At this time, the coordinates (xp, yp, zp) of the point P (X, Y, Z) on the road surface in the camera coordinate system are expressed by the following equations.

   Assuming that the coordinate system of the camera image plane separated from the camera installation position O by the focal length f is (x, y), the coordinate (x, y) on the image plane is obtained as follows.

    Here, the coefficient aij (i = 1, 2, 3, j = 1, 2, 3) is expressed by the following equation.

  The above expression for x, y can be rewritten as the following formula to obtain the local coordinates (X, Y) given the coordinates (x, y) on the image plane and the local altitude Z:

From the above equation, if camera height (X ₀ , Y ₀ , Z ₀ ), camera tilt (ω, φ, κ), and focal length f are given as camera field angle parameters, It can be seen that the coordinates can be associated with the local coordinates.
These seven angle-of-view parameters are defined as unknown variables, and some points are associated with the coordinates (x, y) and local coordinates (X, Y, Z) on the image plane. By solving, seven parameters can be obtained.
The measuring device 4 captures an image through the video cable 3 and stores it in the image memory (step S2). The image capture interval is, for example, 30 images per second.

Based on the image stored in the image memory 43, the image processing unit 41 binarizes an area where the vehicle is likely to exist and an area where the vehicle is not likely to exist in the screen by difference processing (step S3).
In this binarization method, for example, a road surface image (hereinafter referred to as a background image) when no vehicle is present is created in advance, and the brightness difference from the captured image (hereinafter referred to as an input image) is calculated. A pixel that is calculated and a change greater than or equal to the threshold is detected is set to “binarization result = 1”. In the case of a change below the threshold, “binarization result = 0” is set.

Alternatively, for each pixel, the difference in brightness between the previous time and the current time is calculated, and a pixel in which a change greater than or equal to the threshold is detected is set to “binarization result = 1”. In the case of a change below the threshold, “binarization result = 0” is set.
A vehicle candidate is extracted from the binarization result detected by the above method (step S4). This is performed, for example, by the following method.

A cluster of pixels whose binarization result is “1” is searched. A portion where “1” is fixed by a certain number or more is detected and set as a “vehicle candidate”.
However, multiple candidates that are close in distance are combined into one. In addition, if the area is less than a certain value, it is determined as noise and is not regarded as a “vehicle candidate”.
The following position information is stored for each “vehicle candidate”.

・ Vehicle direction of travel (directly under the camera to the tail)
・ Vehicle left end position (left end of measurement range to left end of vehicle)
・ Right end position of vehicle (left end of measurement range to right end of vehicle)
By converting the vehicle position coordinates on the screen thus obtained using the parameters given in step S1, the vehicle position coordinates in the real space can be obtained. If the road is a curved section, the X and Y coordinates may be further transformed so that the Y axis is redefined along the lane curve.

  The image processing unit 41 formats and stores the vehicle trajectory data including the vehicle position coordinates obtained as described above as shown in Table 1. (This data is called “behavior data”.)

In Table 1, for each vehicle ID (for example, 0 to 255 cyclic), the coordinate position of the horizontal line representing the stern and the vehicle head is stored for each day and time. That is, as the “Y coordinate” in Table 1, the value of the coordinate Y along the road traveling direction is stored as described above. As the “X coordinate”, the value of the coordinate X of the center position of the lateral width (width along the X direction) of the vehicle presence area is stored.
Next, collision risk index calculation processing by the event determination unit 42 will be described.

The event determination unit 42 calculates, for each vehicle, a collision risk index (collision risk) with a nearby vehicle (front, rear, adjacent lane, etc.).
For this purpose, the event determination unit 42 collects behavior data for each vehicle as shown in the flowchart of FIG. 5 (step T1).
Next, data correction is performed based on the collected behavior data (step T2). This correction process will be described below.

When the coordinates of the vehicle position are converted from the coordinates on the screen to the coordinates in the real space, the dimension increases by one, so it is necessary to give some value as a constant for the Z coordinate in the real space. When shooting from the back of the vehicle, the car tail position actually captures the bottom of the chassis, but the height of the bottom of the chassis is quite close to the road surface, so there is almost no error in the calculated coordinate position even when the height is zero. Absent.
As shown in FIG. 6, the vehicle head position does not represent the actual front end of the vehicle because the vehicle roof is actually often captured. For this reason, no matter how much the height is given, an error (indicated by α in FIG. 6) is generated in the vehicle head position.

  Therefore, as a countermeasure, as shown in FIG. 7, a method of outputting the vehicle head position by giving the vehicle length by a constant γ when the vehicle tail position is measured is conceivable. In this method, as shown in FIG. When C2 appears to overlap, the position of the vehicle tail of the vehicle C2 cannot be measured, and the position ahead of the vehicle tail by the vehicle length is the vehicle head. For this reason, the inter-vehicle distance is calculated ignoring the vehicle C2, and not only the collision risk of the vehicles C1 and C2 is determined but also the collision risk of the vehicles C2 and C3 is not determined. .

  Therefore, as the vehicle head position output method of the present invention, as shown in FIG. 9, the vehicle head position coordinates on the screen are given height 0 and converted to coordinates in real space. The vehicle head position coordinates thus obtained are the Y coordinates of the point where the tip of the vehicle roof is projected obliquely to the ground. Since this position is clearly ahead of the head position of the vehicle, the Y coordinate is corrected by subtracting the diagonally shaded distance given by a constant β.

Since the value of the constant β depends on the height of the roof of the vehicle, as described above, two-way communication with the vehicle is performed to acquire the vehicle height data, and the height data can be used. It is also possible to obtain the lateral width (width along the X direction) of the vehicle presence area and set the height corresponding thereto. Further, the height may correspond to the width of the vehicle, and the height may be regarded as constant regardless of the vehicle.
Thus, if the Y coordinate is corrected, even in the case of FIG. 8, the distance between the vehicle C2 and the vehicle C1 cannot be measured, so the collision risk of the vehicles C1 and C2 cannot be obtained. Since the position error can be reduced, the inter-vehicle distance between the vehicle C2 and the vehicle C3 can be calculated accordingly, and the collision risk of the vehicles C2 and C3 can be obtained. That is, even in the case of FIG. 8, it is possible to prevent the system from being largely mistaken.

Next, the missing data is complemented (FIG. 5; step T3). As this complementing method, for example, a method of evenly assigning the difference from the data before the loss when tracking is resumed can be considered.
Next, the respective speeds and inter-vehicle distances of the vehicle of interest and the preceding vehicle are obtained as a function of time (steps T4 and T5).
The speed of the vehicle can be obtained by time differentiation of the position coordinates of the vehicle.

However, other than this method, for example, an ultrasonic Doppler radar can be installed in the camera 2 to measure the speed of the vehicle. It is also possible to install two embedded vehicle detectors on the road and measure the time that the vehicle has passed through these vehicle detectors to know the vehicle speed.
The inter-vehicle distance is obtained as a difference between the position of the head of the vehicle of interest and the position of the rear of the vehicle ahead. In this case, as described above, the position of the vehicle head is a position where correction equivalent to the amount of leaning β is performed.

The speed of the preceding vehicle at time t thus obtained is V1 (t), the speed of the vehicle of interest at time t is V2 (t), and the inter-vehicle distance at time t is s0 (t). .
Next, a collision risk index is calculated (step T6).
As the collision risk index, PICUD (Possibility Index for Collision with Urgent Deceleration) proposed in Non-Patent Document 1 can be used. The definition formula of PICUD is as follows.

In this definition, Δt is a reaction delay time until the driver of the following vehicle notices when the behavior of the preceding vehicle changes, and is given as a constant. a1 is the acceleration (<0) when the vehicle ahead decelerates and is given as a constant. a2 is an acceleration (<0) at the time of deceleration of the following vehicle, and is given as a constant.
The first term of this definition formula represents the travel distance from when the preceding vehicle starts sudden deceleration until it stops. The third term is the free running distance until the subsequent vehicle notices the deceleration of the preceding vehicle. The fourth term is a travel distance from when the subsequent vehicle starts sudden deceleration until it stops.

If this PICUD exceeds 0, danger can be avoided. If PICUD is 0 or less, a collision is expected. If PICUD exceeds 0 but falls below a certain threshold, it can be determined that the degree of risk is high.
Therefore, the event determination unit 42 accumulates data for vehicles whose PICUD is below a certain threshold value and uses the data for preventing traffic accidents.

The method of utilizing this data will be described below.
(1) Accumulate trajectory data for each vehicle including PICUD values. In addition to the trajectory data format listed in Table 1, the data format to be accumulated is accumulated as data obtained by adding the preceding vehicle ID, the inter-vehicle distance to the preceding vehicle, and the PICUD value. Based on this accumulated data, for example, if it can be determined that the distance between vehicles tends to be narrow on average in a certain section of the road, measures such as improving the prospect of the front by installing road equipment are taken. be able to.

  (2) When there is a vehicle whose PICUD is below a certain threshold, the event determination unit 42 immediately notifies the traffic control center and transmits the input video. As shown in FIG. 10, this information flow is indicated by an arrow from the event determination unit 42 to the monitoring monitor of the traffic control center. At the traffic control center, the video of the camera in which the vehicle is shown (or close to it) is automatically displayed.

With this system configuration, it is possible to promptly notify the surveillance staff who are monitoring roads with CCTV not only “after an accident has occurred” but “a situation where an accident is likely to occur” and support the monitoring work.
(3) Find a situation “probably an accident” instead of “after an accident”, and alert the attention vehicle and subsequent vehicles. One of the following may be considered as a method of alerting.

(3-1: Use information board)
A road information board (FIG. 11) is installed about 100 m downstream from the camera, and is connected to the measuring device 4 through a communication line.
When a vehicle that is determined to be dangerous by the PICUD value is found, a warning message is displayed on the information board, "Please increase the distance between vehicles." Based on the result of the image processing, the arrival time near the information board is predicted based on the speed of the vehicle, and the display timing is determined so as to be visible only to the vehicle of interest and its subsequent vehicles.

(3-2: Communication alert)
Road beacons (FIG. 11) such as optical beacons and radio wave beacons are installed on the road and connected to the measuring device 4 through a communication line.
When the vehicle is equipped with communication means (onboard equipment) such as an optical beacon, a radio beacon, and a wireless LAN, alerting is performed by communicating with the vehicle of interest and its subsequent vehicles. The timing to transmit to the vehicle is, for example, the vicinity of the beacon of the subsequent vehicle calculated based on the speed of the subsequent vehicle after the predicted arrival time of the target vehicle near the beacon calculated based on the speed of the target vehicle. Estimated arrival time. Thereby, it is possible to warn the driver of the vehicle to increase the inter-vehicle distance.

(3-3: Vehicle control by communication)
When the vehicle has the above-described communication means and can perform automatic control such as braking, automatic braking or the like is performed by communication with the vehicle of interest and its subsequent vehicle.
By the above measures, when there is a driving situation where the speed is high and the inter-vehicle distance is narrow, it can be detected in real time and a traffic accident can be prevented.

  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, in the present embodiment, the vehicle is photographed from behind with the camera, but the traveling vehicle may be photographed from the front. Even in this case, as described above, the collision risk index can be calculated by obtaining the vehicle speed and the inter-vehicle distance. And it is preferable to correct | amend the vehicle tail position of the vehicle which drive | works ahead according to the angle of a camera, and obtain | require an inter-vehicle distance.

It is a road installation figure of the collision risk judgment system of the present invention. 3 is a block diagram illustrating a configuration example of a measuring device 4. FIG. 4 is a flowchart for explaining an image processing method performed by an image processing unit 41. It is a coordinate diagram for defining the coordinates of the real space and the coordinates on the imaging plane. 7 is a flowchart for explaining a collision risk index (collision risk) calculation process by an event determination unit 42; It is a figure for demonstrating the correction method of a vehicle head position. It is a figure for demonstrating the other correction method of a vehicle head position. It is a figure showing the state where vehicles C1 and C2 seem to overlap. It is a figure for demonstrating the other correction method of a vehicle head position. It is a block diagram which shows the other structural example of the measuring device. 6 is a block diagram showing still another configuration example of the measuring device 4. FIG.

Explanation of symbols

2 Camera 3 Video cable 4 Measuring device 41 Image processing unit 42 Event determination unit

Claims (6)

A collision risk determination system comprising a camera for acquiring image data by photographing and a measuring device for determining a collision risk based on the acquired image data,
Vehicle recognition means for recognizing a vehicle traveling on a road based on imaging information of the camera;
Calculating means for calculating the speed and distance between the vehicles when a plurality of vehicles are recognized, and calculating a collision risk index based on the calculated speed and the distance between the vehicles;
Possess and output means for outputting the values of the collision probability index calculated by said calculating means,
The calculating means measures a position 1 on a screen closest to a camera of a vehicle traveling forward among the plurality of vehicles when the plurality of vehicles are obliquely photographed from behind with the camera, and the front is On the road where the position 2 on the screen farthest from the camera of the vehicle following the traveling vehicle is measured, and the position corresponding to the position 2 of the vehicle is obliquely projected to the ground based on the coordinates of the position 2 The position 2 is corrected by subtracting a predetermined distance from the coordinates on the road, and the collision between the position 1 and the corrected position 2 is calculated. Risk assessment system.   The collision risk index includes an inter-vehicle distance, a travel distance from when the preceding vehicle starts to decelerate until it stops, and until a subsequent vehicle of the plurality of vehicles notices the deceleration of the preceding vehicle. The collision risk determination system according to claim 1, wherein the collision risk determination system is calculated based on an idle travel distance and a travel distance from when the subsequent vehicle starts to decelerate until it stops. Claim 1 or claim 2 collision risk determination system according further comprises a storage means for storing a collision risk index calculated by said calculating means. A collision risk determination system according to any one of claims 1 to 3 , and a transmission means for transmitting warning information to a road administrator when a collision risk is determined based on a collision risk index. Including collision risk warning system. A collision including the collision risk determination system according to any one of claims 1 to 3 and a transmission unit that transmits warning information to a vehicle when a collision risk is determined based on a collision risk index. Hazard warning system. 6. The collision risk warning system according to claim 5 , wherein the time when the warning information is transmitted to the vehicle is later than the time when the preceding vehicle passes the installation position of the transmission means.

Images