KR20020085545A - Device and Method for detection of lane violating vehicle using real-time processing of moving images - Google Patents

Abstract

PURPOSE: A system and a method for detecting a lane violation vehicle by using moving pictures are provided to search the lane violation vehicle by processing only partial elements of images. CONSTITUTION: A preprocessing portion(30) receives image data(301). The preprocessing portion(30) recognizes automatically a road luminance value by using the received image data(302). The preprocessing portion(30) processes a recognizing process of a changed value of the road luminance value when the road luminance value is changed(303). An outline processing portion(31) performs a binary process(311). The outline processing portion(31) performs the first spatial axis direction process(312). The outline processing portion(31) performs the first temporal axis direction process(313). The outline processing portion(31) performs the second spatial axis direction process(314). The outline processing portion(31) detects an outline of a vehicle(315). A lane violation vehicle is detected(32).

Description

Detection system of lane violation vehicle by video processing and its method {Device and Method for detection of lane violating vehicle using real-time processing of moving images}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for detecting a lane violation vehicle by moving picture processing, a method and a computer readable recording medium recording a program for realizing the method. Particularly, the present invention relates to a lane in real time using moving picture processing. A system for automatically detecting a violating vehicle, and a method thereof and a computer-readable recording medium having recorded thereon a program for realizing the method.

In order to detect a lane violation vehicle in real time, only 10 to 15 sample points per lane are installed at right angles to the lanes and evenly spaced, and then only pixels of the points are converted to analog (5 or 6 bits). Configure an A / D (Analog / Digital) conversion device. The data from the device contains optical and other noises, so it processes various noises, then outlines the vehicle and detects whether the vehicle violates the lane.

Conventionally, since a lane violation vehicle is detected in such a manner that all the pixels (about 60,000) of the screen are all processed, there is a problem that real-time processing is impossible.

This will be described in more detail as follows.

Ultrasonic or loop vehicle detectors are used as a computer-based traffic information collection system installed in most cities in the world. In particular, many of them are already installed in Korea. These are only measurements at a limited number of points, one to two at each intersection. Therefore, there has been a problem that spatial measurement of multi-lane by one sensor, that is, wide area of measurement, magnification of measurement contents, and promptness of measurement are desired.

On the other hand, applications in the field of measurement for television technology have been used for a long time, and from the viewpoint of the application field, the field of automatic measurement of parameters such as brightness, position, width and area, production process field, medical field, remote sensing field It also extends to a wide range of fields, including biological solids and kinetic analysis of tissues.

Under such phenomena, recently, researches and investigations of traffic flows by moving image processing by combining a solid-state image sensor and a computer have been attempted in the United States, the Netherlands, the United Kingdom, and Japan, but there is a problem in that real-time processing is impossible.

The present invention has been made to solve the above problems, the detection system of the lane violation vehicle by the video processing for detecting the lane violation vehicle in real time through the processing of only some elements, not all elements of the screen and its It is an object of the present invention to provide a computer-readable recording medium storing a method and a program for realizing the method.

1 is a configuration diagram of an embodiment of a detection system of a lane violation vehicle by moving picture processing according to the present invention.

Figure 2 is a detailed configuration diagram of an embodiment of an image information extraction conversion apparatus according to the present invention.

3 is a flowchart illustrating a detection method of a lane violation vehicle by moving picture processing according to the present invention;

4 is a detailed flowchart of an embodiment of an automatic adjustment process of road brightness values according to the present invention;

* Explanation of symbols for the main parts of the drawings

101: video 102: video information extraction converter

103: computer 104: printer

According to an aspect of the present invention, there is provided a system for detecting a lane violation vehicle, comprising: image photographing means for photographing a corresponding road to detect a lane violation vehicle; Receiving image signals obtained through the image capturing means, sampling the image signals at sample points set to a predetermined number at a predetermined position, and quantifying the luminance to extract luminance information of the sample points together with the position information. Image information extracting means; Image information processing means for receiving the luminance information and location information extracted from the image information extraction means to recognize the luminance value of the road, to determine the outline of the vehicle using the luminance value of the vehicle separated from the road, and to detect the lane violation of the vehicle. ; And output means for outputting a detection result of the lane violation vehicle processed by the image information processing means in a predetermined format.

In addition, the present invention, in the lane detection vehicle detection method applied to the detection system of the lane violation vehicle, a predetermined number of predetermined positions are set at a predetermined position instead of processing all the pixels on the screen in order to perform video processing in real time. A first step of extracting luminance information together with position information of an image signal at a sample point; A second step of recognizing a predetermined luminance value consecutively a predetermined number of times for luminance information obtained continuously as a luminance value of a road in a background; A third step of identifying the outline of the vehicle by receiving the extracted luminance information and the location information by dividing the luminance value of the road and the luminance value of the vehicle; And a fourth step of detecting the position of the vehicle based on the obtained vehicle outline and detecting whether the vehicle is in a lane violation.

In addition, the present invention provides a lane violation vehicle detection system having a processor, in which a video signal at a predetermined number of sample points is set at a predetermined position instead of processing all the pixels on the screen in order to perform video processing in real time. A first function of extracting the luminance information with the position information; A second function of recognizing, as luminance values of a road in the background, predetermined luminance values that are continuous for a predetermined number or more with respect to luminance information obtained continuously; A third function of receiving the extracted luminance information and the position information to identify the outline of the vehicle by classifying the luminance value of the road and the luminance value of the vehicle; And a computer-readable recording medium having recorded a program for realizing a fourth function of detecting the position of the vehicle and detecting whether the vehicle is in violation of the lane based on the obtained vehicle outline.

The present invention relates to a system and method for automatically detecting a lane violation vehicle in real time using video processing. Instead of processing pixels on the whole screen to detect lane violations in real time, only 10 to 15 sample points per lane are processed within a maximum of 128 points, and the background (road) is automatically processed. The initial luminance value of is automatically recognized by the computer, and the background luminance value changes according to morning, day and evening so that the computer automatically recognizes this change.

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a configuration diagram of a detection system for a lane violation vehicle by moving picture processing according to the present invention.

In the present invention, a maximum of 128 sample points are used, and 10 to 15 points per lane are uniformly installed at right angles to the lane, and the luminance value of the road (background) and the luminance value of the vehicle (target) are different. By delineating the vehicle, the type of the vehicle is detected, and the lane violation vehicle is detected whether or not the vehicle is on the lane. According to the present invention, in order to detect a lane violation vehicle, a maximum of 128 sample points are provided, and an image information extraction and conversion device for processing only 10 to 15 pixels is configured, and various optical noises are applied to data output from the device. Etc., it processes these noises to outline the vehicle and detects whether the vehicle is in violation of the lane. In addition, since the luminance value changes with the luminance value of the road (background) and time to realize the automation of detection, the computer automatically recognizes this change value.

The system shown in FIG. 1 includes a video camera (or a charge coupled device (CCD) camera) 101, an image information extracting and converting device 102, a computer 103, and a printer 104.

1, which is an embodiment of the system according to the present invention, records the output of a video camera (or CCD camera) 101 directly or in a video recorder, and then processes luminance information with a computer 103. Instead of using the image information extraction and conversion device 102 to process pixels (approximately 60,000 in 8-bit conversion) of the entire screen, real-time processing is performed by handling 10 to 15 luminance information per lane pre-installed on the screen. Make it possible.

The computer 103 processes the received luminance data, accumulates the result, and outputs the result in an appropriate format from the printer 104 after completion of measurement. Here, the printer 104 represents an output device, and may be output not only through the printer but also through other output means such as a monitor.

2 is a detailed configuration diagram of an embodiment of an image information extraction conversion apparatus according to the present invention.

The apparatus shown in FIG. 2 can install as many sample points as necessary in a position required on a screen, and the image information extraction and conversion device 102 of FIG. 1 converts only the sample points to A / D. ), The marker installation processor 201, the synchronization signal extractor 202, the sample point installation and A / D converter 203, the address counter 204, the simultaneous circuit 205, and the buffer memory 206. , Interface 207, controller 208.

The sample point installation and the A / D converter 203 of the image information extraction and conversion device 102 sample up to 128 video signals in pixels on one screen (about 60,000 points) in the first half frame (odd field). (Sampling), the luminance is numerically converted into 5 bits (0 to 31 grades), and temporarily stored in the buffer memory 206. In the latter frame (even field) of the frame 1/30, luminance information is transmitted from the image information extraction and conversion device 102 to the computer 103.

The X and Y coordinates on the screen are displayed by the address counter 204. The X-axis and Y-axis counters are reset to horizontal and vertical sync signals by the sync signal extractor 202, respectively. The X-axis counter of the address counter 204 counts the built-in clock pulse generator output, and the Y-axis counter counts the rounding output of the X-axis counter. When the output of this device is transmitted to the computer 103, an interrupt is generated with a vertical synchronizing signal corresponding to the start of the even field, and the parallel transmission is displayed in units of "Word" through the program channel.

3 is a flowchart illustrating a method for detecting a lane violation vehicle by moving picture processing according to the present invention.

FIG. 3 illustrates a method according to the present invention in which a computer receives an output of an image information extracting and converting device 102 as an input, outlines a vehicle through various processes, and then detects a vehicle in violation of a lane at the position of the vehicle. One embodiment flow chart. The computer 103 of the system according to the present invention includes a preprocessor 30, a contour processor 31, and a lane violation vehicle and type processor 32.

The preprocessor 30 processes an image data input process 301, an automatic recognition process 302 of road brightness values, and a recognition process 303 of change values when the road brightness values change.

The automatic recognition process of the road luminance value 302 may separate the vehicle from the road by knowing the luminance value of the background part (road), which is a reference of detection, so that the computer 103 may automatically recognize (recognize) the luminance value of the road. It is. Since the image of the video camera 101 is 30 frames per second, the fluctuation of the luminance value of the road portion is not abrupt. Therefore, if the input luminance value is continuously continuous, the value is recognized as the road value.

First, the luminance value of the first frame is stored in the memory. Next, the luminance value of the second frame is compared with the first luminance value, and when the value is different, the new value is replaced with the memory. If the values are the same, the number of consecutive times is counted. If the values are consecutive for R number, the value is determined as the luminance value of the road.

When the luminance value data is obtained after the luminance value of the road is set, the automatic recognition process 302 of the road luminance value is jumped to go through the automatic luminance value adjustment process 303.

In the experiment of one embodiment of the present invention, the luminance value of the road was set, and R was set to 3 (1 / 10th). The automatic adjustment process of the luminance value 303 recognizes the change value when the road luminance value changes.

If for some reason the luminance value of the background portion (the luminance value of the road) changes, it is impossible to distinguish between the background portion and the portion other than the background portion (vehicle). If this happens, subsequent measurements are impossible. The luminance value of the background portion changes not only with the change of the luminance value due to the sudden change of the climate, but also with time such as morning, day, and evening. Therefore, the automatic adjustment process of the luminance value 303, which is a method that can automatically follow the change even when the luminance value of the background portion is changed by an external image, will be described in detail with reference to FIG. 4.

As a result, even if the luminance value of the background part changes with time, the computer automatically recognizes the luminance value of the road, and then the road and the non-road part are regarded as the vehicle, and the part corresponding to the road is 0 and the non-road part is the vehicle. In step 2, the process proceeds to the binarization process 311, which is replaced with 1.

In the contour processing unit 31 of the computer 103, the binarization process 311, the space axis direction processing 1 process 312, the time axis processing process 313, the space axis direction processing 2 process 314, and the detection of the vehicle contour Process 315, where the vehicle portion is detected from the image data, which includes optical and electrical noise, thereby processing the noise to detect the perfect vehicle.

In the pattern in which the vehicle part separated from the background part by the binarization process 311 is 1 and the background part is 0, the following phenomenon appears due to electrical and optical noise.

1. There may be noise in the vehicle that is marked as zero.

2. There may be a noise of 1 in the background.

3. Since the brightness between the front and rear parts of the vehicle is similar to the brightness value of the background part, the middle part of the vehicle may be 0, and one vehicle may be two.

The contour processing unit aims to eliminate the problems caused by the above-described phenomenon, and to accurately contour the vehicle, and to do so, inserts the processing in the space axis direction and the time axis direction. After these processes, different vehicle outlines are drawn (312-315) according to the size of the vehicle.

The detection of the lane violation vehicle 32 through the lane violation vehicle and the type processing unit detects the type of the vehicle such as a bus or a passenger car according to the size of the vehicle outline obtained through the detection process 315 of the vehicle outline, and the vehicle. The location of the lane is detected so that the vehicle can be detected.

4 is a detailed flowchart illustrating an embodiment of an automatic adjustment process of road brightness values according to the present invention.

With respect to the road luminance value X (I) set by the automatic recognition process of the road luminance value 302, when the luminance value X (I) ± 1 determined by the background portion is continuous in T1 frames (403, 404), the road luminance value is Change it to a value.

Also, if the road luminance value is significantly different from that of the background part (e.g., sudden change of diary, shadow occurrence, etc.) or larger change is caused by some cause, X (I In the case where ± 2 (402, 405) and X (I) ± 3 (401, 406) are consecutive T2 and T3 times, the road luminance values are corrected.

As a parameter determination method, T1, T2, and T3 are selected as 7 (7/30 seconds), 10 (10/30 seconds), and 20 (20/30 seconds), respectively, by experimental tests. Of course, these values may change depending on the situation.

The present invention described above is not limited to the above-described embodiments and the accompanying drawings, and various substitutions, modifications, and changes are possible in the art without departing from the technical spirit of the present invention. It will be apparent to those of ordinary knowledge.

The present invention as described above enables real-time processing by processing only a part of the screen instead of processing the entire screen using the processing of the video, and can automatically detect the vehicle and the type of the vehicle that violate the lane. It works.

In addition, the present invention automatically checks the luminance value of the road by checking the number of times that the luminance value has the same value, and automatically adjusts the luminance value of the road that changes every time by continuously checking the change in the luminance value with respect to the recognized luminance value of the road. By distinguishing the road and the vehicle in the background with the luminance value, it is possible to accurately detect the lane violation vehicle.

Claims (10)

In a system for detecting a lane violation vehicle, Image capturing means for photographing a corresponding road to detect a lane violation vehicle; Receiving image signals obtained through the image capturing means, sampling the image signals at sample points set to a predetermined number at a predetermined position, and quantifying the luminance to extract luminance information of the sample points together with the position information. Image information extracting means; Image information processing means for receiving the luminance information and location information extracted from the image information extraction means to recognize the luminance value of the road, to determine the outline of the vehicle using the luminance value of the vehicle separated from the road, and to detect the lane violation of the vehicle. ; And Output means for outputting a detection result of the lane violation vehicle processed by the image information processing means in a predetermined format; Lane detection vehicle detection system by the video processing including a. The method of claim 1, The video information extraction means, Synchronizing signal extracting means for continuously receiving the image signal photographed by the image capturing means and extracting a synchronizing signal in real time; Sample point installation and digital conversion means for sampling the video signal for a predetermined number of sample points preset in one screen with respect to the video signal input continuously and dividing and digitizing the luminance of the video signal; And Address counting means for acquiring on-screen coordinates of the sample point using the synchronization signal obtained through the synchronization signal extracting means Lane detection vehicle detection system by the video processing including a. The method according to claim 1 or 2, The video information processing means, Preprocessing means for receiving image information obtained at a sample point from the image information extracting means and recognizing a luminance value of a corresponding road; Vehicle contour detection means for detecting a vehicle contour by separating a road from a vehicle by checking a luminance value of a vehicle distinguished from the vehicle luminance value based on the road luminance value recognized by the preprocessing means, and removing noise through spatial and time axis processing; And Lane violation detection means for checking the type of vehicle, such as a bus or a passenger car, according to the size of the vehicle contour obtained through the vehicle contour detection means, and detecting the position violation based on the position of the vehicle. Lane detection vehicle detection system by the video processing including a. In the detection method of a lane violation vehicle applied to a detection system of a lane violation vehicle, A first step of extracting luminance information together with position information of an image signal at a sample point set by a predetermined number at a predetermined position instead of processing all pixels on the screen to perform a moving image processing in real time; A second step of recognizing a predetermined luminance value consecutively a predetermined number of times for luminance information obtained continuously as a luminance value of a road in a background; A third step of identifying the outline of the vehicle by receiving the extracted luminance information and the location information by dividing the luminance value of the road and the luminance value of the vehicle; And A fourth step of detecting the position of the vehicle by detecting the position of the vehicle based on the obtained vehicle outline; Lane detection vehicle detection method by a video processing including a. The method of claim 4, wherein The first step is, A fifth step of installing a predetermined number of sample points per one lane at regular intervals at right angles to the lane in order to detect a road violation vehicle in real time by receiving an image signal photographing a road; And A sixth step of extracting luminance information and position information by converting analog to digital for the pixel of the sample point; Lane detection vehicle detection method by a video processing including a. The method of claim 4, wherein The second step, A fifth step of recognizing the value as the initial luminance value of the road if the luminance value inputted in recognizing the luminance value of the road is consistently continued in order to detect the lane violation of the vehicle by separating the vehicle from the road; And A sixth step of recognizing the road luminance value by checking whether the luminance value estimated by the change of the road luminance value differs from the road luminance value and checking whether the estimated luminance value is continuous for a predetermined time while recognizing the luminance value of the road. step Lane detection vehicle detection method by a video processing including a. The method of claim 6, The fifth step, A seventh step of accumulating in the memory the luminance value of the first frame of the luminance pattern input to recognize the initial luminance value of the road in the background; An eighth step of receiving a luminance value of a next frame and comparing the luminance value with the luminance value stored in the memory; A ninth step of accumulating the input luminance value in the memory and repeating the operation from the sixth step if the luminance value stored in the memory and the luminance value of the input frame are different as a result of the comparison in the eighth step; A tenth step of counting the number of times that the luminance value stored in the memory and the luminance value of the received frame are equal if the luminance value stored in the memory is the same as the luminance value of the received frame; And An eleventh step of recognizing the initial luminance value of the road for the same luminance value consecutively more than a predetermined number of times Lane detection vehicle detection method by a video processing including a. The method of claim 6, The sixth step, A seventh step of checking a difference value between the recognized road brightness value and the brightness value determined as the background part; An eighth step of determining a predetermined time at which the background luminance value for determining the road luminance value should be continued based on the identified difference value of the luminance value; A ninth step of judging whether or not the luminance values of the background parts have been continued for the predetermined time; As a result of the determination of the ninth step, if the luminance value of the background portion is not continuous for the predetermined time, the tenth step of recognizing the existing road luminance value as the road luminance value continuously, recognizing the luminance value of the background portion, and repeating from the seventh step ; And An eleventh step of recognizing the luminance value of the background portion as the road luminance value if the luminance value of the background portion is continued for the predetermined time as a result of the determination of the ninth stage; Lane detection vehicle detection method by a video processing including a. The method according to any one of claims 4 to 8, The third step, A twelfth step of receiving the extracted luminance information and the location information and determining a luminance value different from the luminance value of the road as the luminance value of the vehicle; A thirteenth step of removing optical and electrical generated noise through spatial axis direction processing and time axis processing; And Fourteenth step of detecting the outline of the vehicle with the luminance value of the vehicle without noise Lane detection vehicle detection method by a video processing including a. In a lane violation vehicle detection system having a processor, A first function of extracting luminance information together with position information of an image signal at a sample point set by a predetermined number at a predetermined position instead of processing all pixels on the screen in order to perform moving image processing in real time; A second function of recognizing, as luminance values of a road in the background, predetermined luminance values that are continuous for a predetermined number or more with respect to luminance information obtained continuously; A third function of receiving the extracted luminance information and the position information to identify the outline of the vehicle by classifying the luminance value of the road and the luminance value of the vehicle; And A fourth function of detecting the position of the vehicle by detecting the position of the vehicle on the basis of the obtained vehicle contour A computer-readable recording medium having recorded thereon a program for realizing this.