Implementation of Expressway Video Super Speed ​​Control System

xiaoxiao2021-03-17  190

Implementation of Expressway Video Super Speed ​​Control System

Source: Application of electronic technology: Sun Huayan TANG Liming Li Yingchun

Abstract: The hardware composition, software function and moving target detection, dynamic target tracking and speed measurement, automatic positioning of license plate characters, and license plate characters automatic identification are introduced. The system can be widely used in highway management, bayonet management, patrol duty, escape vehicle arrest, etc., has a good application prospect. Keywords: video surveillance image processing highway speed monitoring dynamic detection With the rapid development of the national economy, my country's highway has established a change in earth. There are more and more vehicles driving on the highway, and the speed is also getting faster, and the cases related to vehicle traffic have also been continuously rising, and there is a case in cases such as transportation. How to use scientific means to help the public security department effectively control the speed of speed violations on the highway, arresting the escaping vehicles, has become an urgent problem in the public security transportation department. The mature system that is currently capable of complete overspeed monitoring is: microwave radar and laser-based overspeed monitoring system. When it passes the vehicle, the frequency change of the reflected wave is used to monitor the vehicle information, and the oversized vehicle type, license plate number, and other comprehensive traffic information cannot be performed in a timely manner. This highway video speed monitoring system utilizes video image processing technology, non-contact monitoring on the highway lane, obtains operation status information such as speed vehicle speed, license plate number, illegal photo, etc., can be applied to highway management, escape vehicles Catch and other occasions. 1 System Structure The hardware structure of the speed monitoring system is shown in Figure 1. It consists of a portion of the overspeed monitoring camera and a field computer. Install monitoring cameras on the highway (a color panoramic camera, N motorway color cameras) and speed monitoring computers, 24-hour real-time monitoring of all vehicle speed information through vehicles on the highway. The system software includes overspeed vehicle detection and automatic license plate identification. The speed monitoring computer first collects the panoramic image of the highway through the real-time video capture card, and uses the panoramic image to perform overspeed vehicle detection; if illegal vehicles are detected, the camera works to start the corresponding lane, collect the close image and use the profile image automatic license plate identification, The identification result can be divided into the license plate number character, license plate number photo, car illegal photos separately to overtile illegal vehicle database, for delivery; if necessary, automatically automatically to the highway toll station through wireless, cable or fiber optic communication network Processing the server to transfer the license plate number of the violation vehicle, illegal photo information, and perform violations in real time. Software system function block diagram is shown in Figure 2. 2 Software Function 2.1 Moving Target Split [1] [2] Ideally, when performing overspeed vehicle detection from a video image, the front and rear two frame images can be directly compared, and the stationary area, retain the moving area, Determine whether there is a car in the field of view, judge the movement trajectory and speed of the car. However, in the actual imaging process, many factors in the scene include illumination, the number of objects and physical properties of the object in the scene (especially the reflection properties of the surface), the characteristics of the imaging system, and the light source, the imaging system. The spatial relationship between the rooms is all integrated into a gray value of the pixel point in a single image; due to the strong radiation of the space, the change in light changes, and the optical characteristics of the sensor itself, strong interference is generated in each frame image. And noise. Therefore, pre-processing such as an averaging method based on an averaging method based on the image check before the image check is performed, and the real-time background update based on Kalman filtering [3] is pre-processed; then the extraction function [4] is used to split the target with the background. Set CK = {CK (x, y)} represents the current image, RK = {RK (x, y)} represents the reference image, where (x, y) is the coordinate of the pixel point, CK (X, Y) ≥ 0, RK (x, y) ≤ 255, the extraction function EK = (CK (x, y), RK (x, y)) is defined as follows: After finding the matching point, the image is used to use both parallax and the image-calibrated field of view The minimum distance from the minimum resolution and the interval of the image acquisition, the target speed can be calculated, and the target speed is predicted, and the target is predicted and the determination is overspeed.

2.4 Character Automatic Identification [8] Easy to know, 0 ≤ A (CK (X, Y), RK (X, Y)) ≤ 1 in the formula. The image of the actual motion vehicle collected on the highway is detected by the extraction function, and the experimental structure is shown in Figure 3. 2.2 Moving Target Tracking and Speed ​​Measurement [5] While the target tracking, it is necessary to determine the motion speed of the calculated target, so the target tracking is used to use the characteristic point to calculate the vehicle speed. Its point is: Select a set of challenges with invariant properties in the active target window of a frame image, match the same feature point in the next frame image, resulting in parallax. This is the method of feature point matching. The Moravac [6] operator is used as a point feature extraction operator. It is based on an ideal feature point that has a large variance in all directions in all directions. The steps of the feature point extraction are: First, in the window calculation formula of 5 х5, i = N-2, ..., N 2; J = M, M, N is the window The centralized line, the column sequence, Gij is the gradation value of the image at (i, j). Then, it is determined that the alternative feature point is determined. If the advantageous value of the cell is greater than the experience threshold, the image element is an alternative feature point; otherwise, the image element is not a feature point. Finally, the feature point is determined using a method of suppressing local non-maximum M values. Check if each alternative feature point is a maximum of a certain size (5 х5, 7х7, 9х9) window. If there are several alternative feature points in the window, the M value is used as a feature point. The rest are removed. In order to ensure the correct rate of matching, the absolute value of the covariance is the absolute value of the difference, and the minimum value is minimized as a double criterion, determines the matching point to take comfortable to enhance the reliability of the matching results. 2.3 License plate automatic positioning license plate automatic positioning is the first step of automatic identification of license plates, correct and reliable detection of license plate area is the key to ensuring the license plate identification rate. There are currently many license plate automatic positioning algorithms, such as Hough transformations to detect straight lines to extract the license plate boundary area, use grayscale segmentation and regional growth to use texture characteristic analysis technology, etc. However, when actually use, it is difficult to achieve practical requirements. The method adopted herein is: First extract the binary edge image of the vehicle with the Prewitt operator, then use the method of mathematical morphology, color search, and perform automotive license positioning. Figure 4 is a schematic diagram showing the results of the automobile license. The prewitt operator is defined as shown in Figure 5. Due to the specific horizontal texture characteristics of the car image, the license character is most portrait, so the directionality of the previce edge detection operator [7] can be separated from the car background image by the longitudinal edge of the image. The binary edge image obtained by the PRWITT operator is further adopted to generate a communication area image in the mathematical morphology method. The choice of structural elements S is critical to the formation of the candy to pass the formation of the license area, which is very easy to appear with other textured phenomena due to the expansion area, thereby bringing difficulties to further license area extraction. The structural element S employed is a horizontal line segment having a horizontal expansion capacity. After expanding, a plurality of candisted license areas can be obtained, as shown in Fig. 4 (c). Aiming at the area of ​​the license plate, the selection of the license plate, through the size, long-width ratio, Hou Selected car brand character edge density, and then fully utilize the color characteristics of the license plate number area, by searching the number area floor color block And the number color method, further from the image to delete the false license area, get the area where the license plate can be obtained. The graphics are automatically identified for the license plate image after the division, and the license plate image is first binaryized, character segmentation and normalization, and then according to the characteristic library, complete the license plate characters automatically identify. Figure 6 is a schematic diagram of the effect of the character automatic identification.

转载请注明原文地址:https://www.9cbs.com/read-129580.html

New Post(0)