JPH11328578A - Method for recognizing vehicle going to stop at emergency parking zone - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for recognizing a vehicle which enables automatic discrimination of a vehicle going to stop at an emergency parking zone in a tunnel. SOLUTION: A luminance distribution varying slowly in time is taken in a background image of the emergency parking zone from an image of a monitoring camera to photograph the emergency parking zone in the tunnel (St 7) and existence of the vehicle is discriminated by detecting disturbance in the luminance distribution equivalent to the width of the vehicle from a differential image between the image of the monitoring camera and the background image (St 4). Patterns of dirt and water accumulated in the emergency parking zone are successively taken in as the background images, in addition, when the vehicle stops in the emergency parking zone, difference between peaks of the luminance distribution is compared with the width of the vehicle and the vehicle is detected.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a stopped vehicle recognition method for automatically identifying when a vehicle stops in an emergency parking zone in a tunnel, and more particularly to a method for detecting a vehicle with high accuracy. is there.

[0002]

2. Description of the Related Art In a long tunnel, an emergency stop zone 22 is provided in the middle of the tunnel, as shown in FIG. When the vehicle stops here, it is often the case that the vehicle breaks down or an accident occurs, or a fire occurs in a tunnel, and the emergency stop installed in or near the emergency stop zone 22 is used. In most cases, the telephone 23 is stopped for use.

For this reason, the emergency stop zone 22 has a surveillance camera 24
The center is provided with a system that allows the user to visually check the image of the camera and determine whether there is a vehicle stopped in the emergency stop zone 22.

[0004]

However, since such an emergency rarely occurs in a tunnel, it is impossible for a person to continuously observe the image of the surveillance camera 24. Therefore, at present, This video is used to confirm when an emergency call is received.

The technique of automatically detecting a vehicle by analyzing an image is widely used for a traveling vehicle on a main line.
If you try to apply it to the emergency stop zone 22 of the tunnel,
There are the following problems.

[0006] The surveillance camera 24 of the emergency stop belt 22 has a narrow field of view because it is installed for the purpose of displaying only the emergency stop belt 22. If you have a wide field of view like a surveillance camera on the main line,
Vehicles can be easily detected by extracting moving objects in the image, but if the field of view is narrow, the vehicle will suddenly enter and leave the field of view, so detection will not be possible. difficult.

Further, since the emergency stop zone 22 of the tunnel is rarely cleaned, dust accumulates on the road surface and water accumulates on rainy days. Since the pattern of dust and water changes with time, there is a possibility that the pattern is mistaken for the movement of the vehicle.

[0008] In addition, paper falling from the vehicle is in an emergency stop zone.
There is a possibility that the air will be collected in the airplane 22 and soar by the wind pressure of the traveling vehicle, and such a movement may be mistaken for the movement of the vehicle.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for recognizing a stopped vehicle capable of solving the above problems and automatically identifying a vehicle stopped in an emergency parking zone in a tunnel.

[0010]

Therefore, according to the present invention, a surveillance camera for photographing an emergency parking zone in a tunnel is used for the following.
The luminance distribution that changes slowly over time is captured in the background image of the emergency parking zone, and from the difference image between the image of the surveillance camera and the background image, the disturbance of the luminance distribution corresponding to the vehicle width is detected to identify the presence of the vehicle. doing.

Therefore, the pattern of dust and water accumulated in the emergency parking zone is sequentially captured as a background image, and when the vehicle stops in the emergency parking zone, the length of the peak of the luminance distribution becomes shorter. The vehicle is detected by comparing with the vehicle width.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The invention according to claim 1 of the present invention is directed to a recognition method for recognizing a vehicle stopping in an emergency parking zone in a tunnel. The brightness distribution, which changes slowly, is captured as the background image of the emergency parking zone, and from the difference image between the image of the surveillance camera and the background image, the disturbance of the brightness distribution corresponding to the vehicle width is detected to identify the presence of the vehicle. This makes it possible to recognize a vehicle that stops in the emergency parking zone in the tunnel with high accuracy.

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

As shown in FIG. 2, a system for implementing the stopped vehicle recognition method of the present invention converts a surveillance camera 11 installed in an emergency parking zone in a tunnel and a video signal of the surveillance camera 11 into a digital signal. A / D converter 12, an image processing unit 13 for extracting a necessary area from the digital image and extracting a difference from the background, and using the processed image to identify the presence or absence of a stopped vehicle in the emergency parking zone. And an image recognition unit 14.

The surveillance camera 11 captures an image of the emergency parking zone in the tunnel. FIG. 3 exemplifies an image reflected by the monitoring camera 11 and its luminance distribution. FIG. 3A is an image capturing the vehicle 41 stopped in the emergency parking zone,
In a brightness distribution 42 along 40, brightness peaks appear on both sides of the vehicle body. This is because both sides of the vehicle body are formed by a curve with a small curvature, so that, regardless of the color of the vehicle body, the illumination reflects and a bright line appears. Therefore,
The distance between the peaks corresponds to the vehicle width.

FIG. 3B is an image of the zebra pattern 43 drawn in the emergency parking zone. A high luminance portion corresponding to the width of the zebra pattern 43 appears in the luminance distribution 44 along the line 40. ing. Fig. 3 (c) shows the emergency parking zone was made wet with water.
It is an image showing 45, and in the luminance distribution 46 along the line 40, the illumination is reflected on the portion 45 of the water wet pattern, and the luminance appears high.

FIG. 3D shows the paper falling in the emergency parking zone.
It is an image capturing 47, and the luminance distribution 48 along the line 40 includes:
A high luminance portion corresponding to the size of the paper 47 appears.

The A / D converter 12 converts an analog video signal captured by the monitoring camera 11 into a digital signal.

The image processing unit 13 and the image recognition unit 14 operate in the order shown in the flowchart of FIG.

Step 1: When the image data is input, the image processing unit 13 selects a digital image to be input at regular time intervals (Δt) as an object to be processed. Is extracted, and a difference from the set background image is extracted. The initial value of the background image is an image captured when there is no vehicle, and is updated in step 7 described later.

Step 3: The image recognition unit 14 compares the size (area) of the difference image from which the background image has been subtracted with a threshold value. The threshold value is set so that the emergency parking zone gets wet with water on a rainy day, and a gradual change to the extent that the pattern spreads is equal to or less than the threshold value. If the size of the difference image is smaller than the threshold, step 7: set the image cut out in step 2 as a new background image. Step 4: If the size of the difference image is larger than the threshold value in Step 3, the length of the difference image itself as shown in FIG. 4A or the difference image as shown in FIG. When the distance to the difference image is compared with the vehicle width or the vehicle length, and a length corresponding to the vehicle width or the vehicle length can be found, that is, the luminance difference corresponding to the vehicle width or the vehicle length is found in the difference image. Step 5: It is recognized that there is a stopped vehicle.

Step 6: If no disturbance in brightness corresponding to the vehicle width or the vehicle length is found in step 4, the process is terminated.

By such a process, the zebra pattern 43 shown in FIG. 3B becomes a background image, and is deleted from the difference image in the process of step 2.

The patterns of dust and water that accumulate in the emergency parking zone shown in FIG. 3C are sequentially captured as background images in the processing of steps 3 and 7.

The papers shown in FIG. 3D are excluded from the recognition of the stopped vehicle by the processing in step 4.

As shown in FIG. 3 (a), when the vehicle stops in the emergency parking zone, at step 4, the peak of the luminance distribution 42 is compared with the vehicle width or the vehicle length. Is done. Therefore, the presence of the vehicle is reliably detected, and there is no detection omission.

By performing processing such as issuing an alarm on the basis of the recognition result in step 5, automatic monitoring of the emergency parking zone becomes possible.

[0028]

As is apparent from the above description, the method for recognizing a stopped vehicle according to the present invention can recognize a vehicle stopped in an emergency parking zone in a tunnel with high accuracy.

Therefore, it is possible to automatically monitor the emergency parking zone in the tunnel using the method for recognizing a stopped vehicle, and it is possible to quickly detect when the vehicle stops in the emergency parking zone. Therefore, it is possible to detect an abnormality at the center more quickly than when a person gets off the vehicle stopped in the emergency parking zone and makes an emergency call, and it is possible to quickly take necessary measures.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a procedure of a stopped vehicle recognition method according to an embodiment;

FIG. 2 is a configuration diagram of a system that implements a stopped vehicle recognition method according to the embodiment;

FIG. 3 is a diagram illustrating an image of a surveillance camera in an emergency parking zone;

FIG. 4 is a diagram for explaining a comparison between disturbance in luminance distribution and vehicle width in the stopped vehicle recognition method according to the embodiment;

FIG. 5 is a diagram showing an emergency parking zone in a tunnel.

[Explanation of symbols]

 11 Surveillance camera 12 A / D conversion unit 13 Image processing unit 14 Image recognition unit 21 Main line 22 Emergency parking zone 23 Emergency telephone 24 Surveillance camera 40 Transverse line of luminance distribution display 41 Vehicle 42, 44, 46, 48 Luminance distribution 43 Zebra pattern 45 Water stain pattern 47 Paper

Claims (1)

[Claims] 1. A recognition method for recognizing a vehicle that stops in an emergency parking zone in a tunnel, the method comprising: detecting, from an image of a surveillance camera photographing the emergency parking zone, a brightness distribution that changes gradually with time in the emergency parking zone. A method for recognizing a vehicle, comprising capturing a background image of a band, detecting a disturbance in a luminance distribution corresponding to a vehicle width from a difference image between the image of the monitoring camera and the background image, and identifying the presence of the vehicle.