Vehicle And Pedestrian Detection Technique Street Surveillance Video

In recent years, with the rapid increase in the number of city vehicles, urban road traffic safety problems increasingly grim. Fast and effective solution to this problem, the video image processing-based video surveillance system has been widely applied to traffic safety monitoring activities.The main contents of this paper include: image preprocessing, image dynamic background modeling and moving target object detection. Wherein the target detection is the main content streetscape image of the video surveillance system. The main work is as follows:(1) Street View image pre-processing and image dynamic background modeling Street: Street View image preprocessing is the main factor based on image blurring, fuzzy image processing, focusing on Wiener filtering methods. Image de-noising process, the salt and pepper noise removal algorithm based on the use of salt and pepper noise adaptive weights. Gaussian noise filtering using image processing partial differential equations. Background extraction is the basis of the image target detection, the extracted background a direct impact on target detection. In this paper, using a modified Gaussian mixture model adaptive algorithm to extract the background model, this algorithm can effectively adapt to changing detection area.(2) Street View image of the vehicle testing: real-time background modeling, most of the area to remove background interference. Street View images for the characteristics of the vehicle, the use of modified binary algorithm to obtain the image of the vehicle model, and then extracting the outline of the vehicle by means of edge detection. Experiments show that, to ensure detection accuracy of the premise, the algorithm can effectively improve the efficiency of the vehicle testing time, and has a strong adaptive ability, the detection precision.(3) Image streetscape pedestrian detection: First detected in the image feature point model pedestrian effective, research contour feature pedestrian and non-pedestrian sample isolated from the detection sample and pedestrian area in the image. Operators then edge detection method for pedestrian sample image edge detection by Canny, combined with morphology for further image processing. Thereby obtaining efficient and accurate pedestrian profile.