JPH0721488A - Traffic flow abnormality monitoring and supporting device - Google Patents

Abstract

PURPOSE:To provide a traffic flow abnormality monitoring and supporting device capable of quickly detecting abnormality in a traffic flow without drop ping accuracy. CONSTITUTION:This traffic flow abnormality monitoring and supporting device is provided with a camera 1 for monitoring a traveling road, a tracking unit 31 for quickly detecting a sudden event by the detection of a speed change, positional movement, etc., based upon picture data picked up by a camera 1 and generating the 1st alarm such as warning, an accumulation unit 32 for accumulating and averaging picture data in each prescribed period, detecting a sudden event by comparing edge images, and outputting the 2nd alarm indicating an alarm, and an integrating part 33 for integrating the 1st and 2nd alarms and outputting the integrated alarm.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a traffic flow abnormality monitoring support device for detecting a traffic flow abnormality such as an accident on a road on which a vehicle travels or the occurrence of traffic congestion.

[0002]

2. Description of the Related Art As a conventional traffic flow abnormality monitoring support device, the movement of a vehicle is extracted and analyzed, for example, in terms of speed and position (travel direction), and the presence or absence of an abnormality such as an accident is discriminated at high speed in a short time. When a situation is detected, an alarm is sent to the host system. In this type of device, when an alarm is transmitted, a higher-level system allows a person to look at the monitor screen to confirm the occurrence of an accident and to display the obtained information, for example, "accident occurrence" on the display board, and I try to let the car know.

[0003]

In the above-mentioned conventional traffic flow abnormality monitoring device, since the alarm is output in a short time after the occurrence of the event, the reliability of the signal is not always high. However, it is possible to make a misjudgment. In particular, when the determination is performed in a short time by image processing using a camera, erroneous recognition may occur due to temporary disturbance of traffic flow or various environments. Originally, an abnormal traffic flow phenomenon such as an accident has a very low occurrence degree, and there are not so many opportunities for a true alarm to be output. On the other hand, there are relatively many false alarms due to malfunctions, and even if an alarm is output, the probability of an accident occurring is low. Particularly in a monitoring support device in which a person intervenes in a higher-level system, when false alarms occur frequently, there is a problem that reliability of the device by a human decreases.

Further, in order to improve the reliability, an advanced algorithm is applied (which requires a long processing time), and the behavior of the vehicle in the case of an accident is observed by a long time unit, whereby the accident It is possible to output a highly reliable alarm by clarifying the distinction between traffic flow abnormal phenomena and temporary traffic flow disturbances and reducing the occurrence of false recognition. There is a problem that it takes time.

The present invention has been made in view of the above problems, and an object thereof is to provide a traffic flow abnormality monitoring support device capable of detecting a traffic flow abnormality in a short time and yet ensuring reliability. I am trying.

[0006]

A traffic flow abnormality monitoring support device according to claim 1 of the present application provides an image pickup means for picking up a predetermined road and an image picked up by the image pickup means. A first abnormal traffic flow measuring device that detects a vehicle speed and / or position based on the vehicle speed and / or position, detects an abnormal event at high speed in a short time based on the vehicle speed and / or vehicle position, and outputs a notification signal to that effect. And, the image data obtained by the image pickup means is accumulated for each pixel, an averaging process is performed, and the detected image data and the model image data are compared,
A second abnormal traffic flow measuring device that detects an abnormal event based on the comparison result and outputs a notification signal indicating that, and an abnormal event occurs upon receiving the notification signal from the first abnormal traffic flow measuring device. Output a first signal that means
And an output means for outputting a second signal which means the occurrence of an abnormal event complementing the first signal in response to a notification signal from the abnormal traffic flow measuring device.

In this traffic flow abnormality monitoring support device, the image pickup means images the road on which the vehicle is traveling at a predetermined cycle, and based on the obtained image data, the first abnormal traffic flow measurement device is used, for example, the vehicle speed and the vehicle. An abnormal event is determined by detecting the position and measuring the moving direction by changing the vehicle speed or detecting the position. When the abnormal event is detected, a notification signal indicating that is output. Further, in the second abnormal traffic flow measuring device, the image data obtained by the image pickup means is accumulated for each pixel, the averaging process is performed, the detected image data and the model image data are compared, and the vehicle is stopped as a result of the comparison. If so, a notification signal indicating the occurrence of an abnormal event is output. In response to the notification signal from the first abnormal traffic flow measuring device, the output means first outputs the first signal indicating the occurrence of abnormality. After that, in response to the notification signal from the second abnormal traffic flow measuring device, the second signal that complements the abnormal occurrence of the first signal is output.

First, the first traffic flow abnormality measuring device detects an abnormality and outputs an abnormality occurrence in a short time, and the second traffic flow abnormality measuring device outputs an abnormal event with high accuracy, although the time is slightly delayed. Since the notification signal indicating the occurrence is output, the lack of accuracy in the first traffic flow abnormality measuring device is complemented by the measurement result of the second traffic flow abnormality measuring device. The traffic flow abnormality monitoring support device according to claim 2 of the claims detects the speed and / or the position of a vehicle based on an image capturing means for capturing an image of a specific road and an image captured by the image capturing means. A first abnormal traffic flow measuring device for detecting an abnormal event at high speed in a short time based on the speed and / or vehicle position, and image data obtained by the image pickup means are accumulated for each pixel and an averaging process is performed. The second that compares the detected image data with the model image data and detects an abnormal event based on the comparison result
Of the abnormal traffic flow measuring device and the signal indicating the degree of abnormality from the first and second abnormal traffic flow measuring devices, and applying a predetermined rule, a signal indicating whether or not an abnormal event occurs by fuzzy reasoning. And fuzzy inference means for outputting.

In this traffic flow abnormality monitoring support device, high-speed abnormal event detection output by detection of vehicle speed, vehicle position, etc. by the first abnormal traffic flow measurement device and accumulation by the second abnormal traffic flow measurement device. High-accuracy abnormal event detection outputs by processing, averaging processing, and comparison processing are input to fuzzy inference means and integrated by fuzzy inference to reduce false alarms.
An optimized signal with a short detection time is generated and output.

[0010]

The present invention will be described in more detail with reference to the following examples. FIG. 1 is a block diagram showing the configuration of a system in which the present invention is implemented. This system consists of multiple cameras 1 _-1 , ..., 1 _-m installed on the right shoulder of the road.
A camera unit 2, a device group 3 installed in a management office, a device group 4 installed in a traffic control center, and a road information board 5 installed at a required location on the road. .

The equipment group 3 of the management office includes cameras 1 _-1 ,
..., and ITV controller 6 for controlling the 1 _-m, camera 1 _-1, ..., the image processing unit _7-1 corresponding to 1 _-m, ..., includes a 7 _-m, respectively traffic flow abnormality monitoring processing , And the image processing device 8 for performing the same, and the distributor 9 _-1 for distributing the image pickup signals of the cameras 1 _-1 , ..., 1 _-m to the ITV control device 6 and the image processing units 7 _-1 , ..., 7 _-m. , ..., 9 _-m and monitor 10 _-1 for check,
... with 10- _n .

The equipment group 4 of the traffic control center is
Report device 11 and camera 1_-1, ... 1 _-mControl from the center
ITV control device 12 for displaying and a display area of road information board 5
Information board control device 13 for controlling the physical processing and monitor 14_-1,
…, 14_-pIs. In this system, the camera
1_-1, Image processing unit 7_-1Focusing on the camera 1, _-1Is an example
For example, as shown in FIG.
It is installed on a pillar 23 that stands on the road side,
Capture the image taken in the period. The captured image data is
Distributor 9_-1, ITV control device 6, ITV control device 12
It is also transmitted to the road traffic information device 11 and the monitor 1
Four_-1Etc., a road surface image can be taken.

On the other hand, the camera 1_-1Image data captured by
Is the distributor 9_-1Image processing unit 7 _-1Is also taken into
The presence / absence of traffic flow abnormality is determined here, and the traffic flow abnormality is detected.
If this happens, an alarm signal indicating that fact is output, and this alarm
The traffic signal goes through the ITV control devices 6 and 12, and
Displayed according to the type of alarm sent to the information device 11.
The road information board 5 is controlled by the control of the road information board control device 13.
Is displayed. Other camera 1_-2, ... 1_-m,Image processing
7_-2, ... 7_-mIs also the same.

The camera unit 2 and the image processing device 8 are essential parts of the traffic flow abnormality monitoring support device of the present invention, and therefore these parts will be further described. Image processing unit 7
As shown in FIG. 3, a tracking unit 31 that takes in image data from the surveillance camera 1 and performs high-speed detection of an unexpected event.
And an accumulating unit 32 that performs highly accurate detection of an unexpected event,
It is composed of an integration unit 33 that determines an unexpected event. When an unexpected event occurs, the tracking unit 31 does not have sufficient accuracy, but a sudden abnormality is detected at high speed in a short time.
The first alarm (notification signal) is output. After that, the accumulating unit 32 performs the accumulating process at regular intervals over a certain period of time to detect an unexpected event with high accuracy. Cumulative unit 32
Second alarm (notification signal) when a sudden event is detected in
Is output. When the first alarm is received from the tracking unit 31, the integration unit 33 first outputs the first alarm, and then, when a second alarm is received after a lapse of time, a second alarm that guarantees higher accuracy than the first alarm is output. Output and as a result
First of all, the accuracy of the first alarm is not sufficient, but if a notification of the occurrence of an unexpected event is issued and the second alarm is output with a slight time delay, a warning of the occurrence of an unexpected event is further provided to complement the first alarm. I do.

The details of the processing in the tracking unit 31 are shown in FIG.
This will be described with reference to the flow chart shown in FIG. First, time t₂
Input image f (t₂) Is taken in (ST1), after Δt
Input image f (t₂+ Δt) (ST2), both images
Image time difference f (t) -f (t₂Calculate the absolute value of + Δt)
(ST3), further horizontal profile (imaging
The image is taken in the direction perpendicular to the road surface. That is
Create a pixel histogram in the vertical direction to the road
It ) (ST4), normalize (ST5), and
Binarization processing is performed with various threshold values (ST6). Regular here
Compensation is the correction for the region where the image is taken in the vicinity on the same screen.
If the value is 1, the correction value for the area that captured a distant image
To max2 and normalize with a linear correction value
The size (length) of the object is corrected by the difference in perspective.
Is Umono. Furthermore, it is binarized, and
After making a correction to connect the parts as one vehicle (S
T7), determine the current position of the vehicle (ST8),
The processes of ST1 to ST8 are repeated until the passage. And time
Tick T = t₁And T = t₁Time T from the two screens at + Δt
= T₁The vehicle position at time T = t₂And T =
t₂From the two screens at + Δt, the time T = t₂In
The vehicle position is calculated (ST9), and this time t₁And t ₂To
Calculate the speed of the target vehicle from the open position (ST10),
The speed at the next time is calculated from the four images as shown in
The change is calculated (ST11). In this case, there are 8 screens
Instead, it is possible to use 6 sheets. In addition, the position information (S
A running position abnormality is detected from the trajectory of (T12) (ST1
3), the degree of speed change, the abnormal position, and the abnormal running degree
Is output (ST14). This abnormal running degree is analog
It is output as a value, but this value is separated by an appropriate threshold value
Therefore, the first alarm signal may be output.

Details of the processing in the accumulating unit 32 will be described with reference to the block diagram shown in FIG. 5 and the flow chart shown in FIG. First, the time variable i is set to 1 (ST20). This variable i is used to repeat the processing from ST21 to ST29 described later N times at a constant cycle. Next, from the surveillance camera 1, the input image f (t) is added to the image accumulator 41 (ST21), and the current image data is accumulated in each pixel data (0 at the reset time) stored in the image memory 42. It is calculated (ST22) and stored in the image memory 42. This accumulation process is performed from t = T ₁ to t = T _N , and t
= T _N and the same process is repeated.

First, at i = 1 (t = T ₁ ), the accumulated data is read from the image memory 42 to the buffer 44, and the averaging unit 45 divides the brightness of each pixel by the number of accumulations to obtain an average value. Is calculated (ST23), and edge detection is performed (ST24). Then, the absolute value of the difference from the background edge image registered in the model edge image storage unit 47 is obtained by the calculation circuit 48 [comparison processing] (ST25), and this is binarized by the binarization circuit 49. When the vehicle does not exist in the image pickup screen, the image in which the edge is detected is an image of only the road surface, so the difference from the background edge is 0, and when binarized, the number of pixels that become 1 is extremely high. Few. However, if the vehicle is stopped,
When there is a large obstacle, the image of that portion is extracted by calculating the absolute value of the difference, and the number of pixels that become “1” when binarized increases. The number of pixels is counted by the counter 50, and the ratio of the number of output pixels to the total number of measured pixels is calculated by the calculation circuit 51 (ST26). Here, the total number of measurement pixels is
The total number of screen pixels, that is, the total number of white pixels and black pixels. For example, if this ratio is equal to or greater than a predetermined value or a value within a predetermined range, it is determined that an abnormal state has occurred, and as an unexpected vehicle presence signal,
That is, the alarm 2 is output and the process proceeds to the next step ST28 (ST27). If no abnormal state is measured, ST27
And skip to ST28. When fuzzy inference which is a second embodiment of the present invention described later is used,
Output as an analog value.

In ST28, it is determined whether i = N (ST)
28), if the determination is NO, increment i by 1.
(ST29), return to ST21, and next timing
T₂The processing of ST21, ..., ST29 in
The accumulation in the accumulation memory 42 is continued. Such processing
Increases i, that is, the time t = T₂, T₃, ..., T_N
It is repeated every time. That is, cumulative processing
Is continued. When i = N, the determination in ST28 is YES.
(At this point, t = T_N), The cumulative memory 42 is reset
(ST30) and returns to ST20, i =
Set to 1. Time is T ₁As the time goes by,
Since the number of times increases, the accuracy improves with the passage of time.
It That is, the accuracy characteristic is t = T_NAfter accumulating up to
Since the product image is reset, it has a sawtooth characteristic.

In this embodiment, when an unexpected event occurs in which the vehicle is stopped due to an accident or the like, first, in the tracking unit 31,
A sudden event is detected at high speed by detecting a change in vehicle speed, a position abnormality, etc., and the first alarm is output. This first
The alarm is output at high speed and in a short time. Next, in the accumulating unit 32, the accumulative data detects a sudden event, and the accuracy increases in the order of time T ₁ , T ₂ , ..., _TN , so that the accuracy of the first alarm is insufficient. The lack can be supplemented in time sequence with the second alarm.

The integrating unit 33 may use the fuzzy reasoning device shown in FIG. In FIG. 7, to the input of the fuzzy inference device 54, a signal S ₁ having a probability of detecting a tracking unit sudden event of −0.5 is passed through a latch circuit 51 and a signal x ₁ is input.
The signal S ₂ of the probability of accumulative unit sudden event detection being −0.5 passes through the latch circuit 52, the signal x ₂ and the signal S _{3 of} cumulative unit cumulative completion screen number / maximum cumulative screen number passes through the latch circuit 53. Add signal x ₃ . The fuzzy inference device 54 is a circuit device for executing fuzzy inference, and a dedicated device, a processor programmed to execute the fuzzy inference, or the like is used. The fuzzy inference device 54 performs fuzzy inference according to the rules stored in the fuzzy rule memory 55, and outputs the probability y ₁ of sudden event detection as an inference result.

The following fuzzy rules are stored in the fuzzy rule memory 56. if (if) x ₁ = PL, x ₂ = PL, x ₃ = PL then (if it is) y ₁ = PL if x ₁ = NL, x ₂ = NL then y ₁ = NL if x ₁ = PM, x ₂ = NL x ₃ = NL theny ₁ = NL Note that PL is "positive and large", PM is "positive and medium",
NL means "negative and large". Accident detection probability of the tracking unit used in this fuzzy rule −
X ₁ signal from 0.5, the membership function of an input of x ₃ signal from x ₂ signal, and the accumulated units cumulative completion screen number / maximum accumulated number of screens from the cumulative units unexpected incident detection certainty -0.5 Are shown in FIGS. 8, 9 and 10.
The probability y ₁ of sudden event detection, which is the output of the fuzzy inference apparatus, is output as an analog value with a degree of reliability. This degree y ₂ may be divided into a predetermined threshold value to serve as a warning signal (second alarm), and since it is an analog value, it may be displayed in a more detailed expression on the display device, or It may be given as processing data.

In this embodiment, the output of the tracking unit and
Since the fuzzy inference is executed using the output of the accumulating unit, and the inference is comprehensively made from the output value of the tracking unit and the output value of the accumulating unit, the second alarm can be output earlier. Since the degree of risk can be calculated based on the degree of accuracy rather than a binary value indicating whether it is highly accurate or dangerous or not, it is possible to detect an unexpected event early and accurately.

[0023]

According to the invention described in claim 1, for example, a notice is displayed on the display means by a high-speed notification signal from the first abnormal traffic flow measuring device in a short time, and the second abnormal traffic flow is further displayed. For example, a warning can be displayed by the highly accurate notification signal from the flow measurement device, so that the traffic flow abnormality can be detected at high speed and with high accuracy.

According to the invention of claim 2, fuzzy reasoning is applied to the outputs of the first abnormal traffic flow measuring device and the second abnormal traffic flow measuring device to obtain the certainty of the sudden event detection. A high-speed and highly accurate traffic flow abnormality monitoring support device can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a system in which the present invention is implemented.

FIG. 2 is a perspective view showing an installation example of the apparatus according to the embodiment of the present invention.

FIG. 3 is a block diagram showing a main configuration of an apparatus according to an embodiment of the present invention.

FIG. 4 is a flowchart for explaining the operation of the tracking unit of the apparatus according to the embodiment.

FIG. 5 is a block diagram showing a functional configuration of an accumulating unit of the apparatus of the embodiment.

FIG. 6 is a flowchart for explaining the operation of the accumulating unit of the apparatus of the embodiment.

FIG. 7 is a block diagram of a fuzzy inference unit showing another embodiment of the integration unit of the above-described embodiment apparatus.

FIG. 8 is a diagram showing a membership function with a certainty of −0.5 of detection of a tracking unit sudden event input to the fuzzy reasoning apparatus of the apparatus of the embodiment.

FIG. 9 is a diagram showing a membership function input to the fuzzy inference apparatus with a probability of accumulative unit sudden event detection of −0.5.

FIG. 10 is a diagram showing a membership function of cumulative unit cumulative completion screen number / maximum cumulative screen number input to the fuzzy inference apparatus.

[Explanation of symbols]

 1 Camera 31 Tracking Unit 32 Cumulative Unit 33 Integration Unit

Claims (2)

[Claims] 1. An image pickup means for picking up an image of a predetermined road, and a vehicle speed and / or position is detected based on an image picked up by the image pickup means. A first abnormal traffic flow measuring device that detects an abnormal event and outputs a notification signal to that effect, and image data obtained by the image pickup means is accumulated for each pixel, and averaging processing is performed. And the model image data are compared with each other, and based on the comparison result, a second abnormal traffic flow measuring device that detects an abnormal event and outputs a notification signal to that effect; and a notification from the first abnormal traffic flow measuring device. Receiving a signal, outputting a first signal that means the occurrence of an abnormal event, and meaning, after outputting, generating an abnormal event that complements the first signal in response to a notification signal from a second abnormal traffic flow measuring device. Output a second signal A traffic flow abnormality monitoring support device, comprising: 2. An image pickup means for picking up an image of a predetermined road, and a speed and / or position of a vehicle are detected based on an image picked up by the image pickup means. The first abnormal traffic flow measuring device for detecting an abnormal event and the image data obtained by the image pickup means are accumulated for each pixel, an averaging process is performed, the detected image data and the model image data are compared, and the comparison result A second abnormal traffic flow measuring device for detecting an abnormal event, and a predetermined rule is applied by receiving a signal indicating the degree of abnormality from the first and second abnormal traffic flow measuring devices, and fuzzy inference is performed. A traffic flow abnormality monitoring support device comprising: a fuzzy inference means that outputs a signal indicating whether or not an abnormal event has occurred.