Research On Moving Object Detection And Tracking Based On Video Images Sequence

Image processing is one of the most popular research directions in computer science. It is widely and importantly applicable to traffic video monitoring, aerospace, robot vision and medical image analysis. So this subject is considered as one of the most development prospective subjects at present. Moving object tracking in video sequence is an important and necessary technology within the filed of computer vision. Key problems of moving target detection and tracking technology are accurate detection and foreground reconstruction, effective tracking. Based on the analysis and research of the key technology, this paper studies on system is divided into the following three parts:Firstly, in the area of moving target detection, two kinds of commonly used moving target detection method are analysed and compared in this paper. Respectively from the working principle, algorithm and simulation experiment three aspects, this paper analyzes and compares the respective advantages and disadvantages.Secondly, in the aspect of moving object tracking, this paper mainly studied the traditional Mean Shift tracking algorithm. The specific research content includes the theory of Mean Shift and its extended form, the choice of color space in Mean Shift algorithm and steps of target tracking algorithm based on Mean Shift, then through the simulation test the moving target tracking effect of Mean Shift algorithm in different cases. The experimental results show that the traditional Mean Shift algorithm has good accuracy and robustness under the condition of slow motion, but inefficiency of algorithm to track and even failure when the target moving too fast or face severe occlusion.Thirdly, to solve the fast moving object tracking and severe occlusion cases failure problem, a new effective approach combining Mean Shift and Kalman filter is proposed, and study step by step through the following three links:I. the modeling process of Kalman filte which takes advantage of its ability of prediction and completed the combination of Mean Shift and Kalman filter is discussed in detail; 2. According to the problem of tracking failure when target severe occlusion, a method to decide the presence of severe occlusion and a corresponding solution is proposed; 3. Showed the effect of the improved algorithm in both tracking cases and analyzed the results through the experiment of simulation many times. Comparing the experimental results of traditional Mean Shift algorithm with the improved algorithm, it proved that the improved algorithm can achieve continuous and stable tracking, the tracking effect has improved significantly compared with traditional Mean Shift algorithm.In order to solve the problem of target tracking failure in fast moving cases, this paper proposed an improved method which combained Mean Shift algorithm with Kalman filter, can effectively solve the problem of fast moving target tracking. For the situation of severe occlusion, a method to decide the presence of severe occlusion and give corresponding processing is proposed.