The Design And Implementation Of Moving Object Detection Tracking System

Image sequence based on moving object tracking is a fundamental problem for computer vision research and has been widely studied. Visual tracking technique has many applications, such as video surveillance, video analysis, video indexing, video based motion analysis and synthesis, motion-based human identification. Visual tracking technique has made great progress in the past several years, but practical experience has shown that visual tracking technologies are currently far from mature. A great number of challenges need to be solved before one can implement a robust visual tracking system for commercial applications.Under the framework of theory of object detecting based on the background difference, this paper proposed some useful improved algorithm about background modeling, threshold segmentation, shadow suppression and object association. The main contributions of this thesis can be concluded as follows.1:Image segmentation is one of the main content on the moving object tracking. This paper compared the existing performance difference of several segmentation algorithms and proposed an adaptive threshold segmentation algorithm about the actual object detection, including the good real-time performance.2:Background difference is one of the most general methods of moving object tracking. Under the actual situation, changing of natural environment will lead to the disturbance of background and light intensity changes. The paper presents a non-parameter estimated background extraction and updating algorithm to solve the problem of background light changes effectively.3:In order to study the traditional moving object tracking, object tracking algorithm is presented based on Kalman filter. Through the establishment of inter-frame relational matrix, the tracking progress is divided into5cases and analyzed. When occlusion occurred between two or more object, we are matched by the template of pre-extraction and obtain accurate objects region. Experimental results show that the algorithm can achieve accurate real-time tracking of moving objects in the situation of complex background. 4:Study a robust visual object tracking method on the base of a sparse approximation in a particle filter framework. We test the algorithm on a large of video sequences, which shows excellent performance involving heavy occlusions, drastic illumination changes.