English Title Of Pedestrian Detection Technolog Of Intelligent Video

With highly developed information society, information society computing vision to become a very important area. According to IDC’s research, the average growth rate of China’s video surveillance market has exceeded 30 percent, China is facing a good development trend in the field of intelligent video surveillance in the next few years will certainly be considerable development. Intelligent Video (IV, Intelligent Video) is the use of a computer powerful data processing capability, drawing on the massive high-speed video analysis to obtain critical information useful to the user, removing irrelevant information.This paper studies analyzing video surveillance to detect moving objects identification and tracking technology, the basic idea is to image sequence processing, and to study the movement of the moving target objects for video image analysis. The main purpose is to detect from video sequences, the segmentation and tracking of human pedestrians, and pedestrian behavior to describe and understand. The key technologies include:image preprocessing enhanced target detection pedestrians, pedestrians object extraction, post-processing. Where the image preprocessing is to reduce noise, highlighting describe objectives, increased sharply contrast; pedestrian target detection is to split the background and foreground, the camera moves can be divided according to whether the pedestrian target detection in static background and pedestrian targets under dynamic background detection; pedestrian target extraction is the removal of the background, foreground region obtained; area after treatment is optimized detection, filtering more noise.This paper describes a pedestrian tracking moving target detection, tracking pedestrians in addition to looking for pedestrians own characteristics, but also find the tracking method pedestrians. Existing target tracking methods are mainly based on tracking feature matching method based on motion prediction and tracking methods such as contour-based tracking method. This paper presents a moving target detection range based distribution model and line coding method based on tracking, and focuses on pedestrian detection and tracking technology. First image sequences to noise and format conversion pretreatment need to deal with a lot of pedestrian detection in video sequences, a large number of images in the presence of noise will affect the identification of the image. Therefore, the identification of pedestrians before the target in the image, the image file first filtering simple space or time to eliminate environmental changes due to noise generated; then converted the image background and foreground extraction interval distribution model based algorithm to obtain accurate moving target location and size; finally segment encoding the moving object tracking method based on moving targets by matching front and rear frames, extract trajectories of moving objects, tracking the trajectory can achieve real-time tracking of moving targets.