Research On Moving Object Detection Andtracking Algorithms In Video

Moving object detection and tracking is a hot topic in the field of computer vision. It is a highly integrated discipline which relates to statistics, cognitive science, signal processing, artificial intelligence, computer science and engineering and many other fields of advanced technology. It has very important practical value in human–computer interaction, intelligent transportation, security monitoring, vision navigation, video image compression and medical diagnosis. In recent years, along with the rapid development of computer and information technology, a large number of researchers and research institutions made a thorough research of moving object detection and tracking, and put forward corresponding solutions according to different problems. Nevertheless, there are still a lot of difficulties in designing universal, effective, and real-time moving object detection and tracking algorithms due to the complex and uncertainty scenarios. Therefore, the researching on moving target detection and tracking methods has great theory significance and utility value.Specifically, this paper researches the moving object detection and tracking algorithms in video. First, the research background and research status of moving target detection and tracking are introduced. Next, the basic knowledge of image preprocessing is discussed. Then, the moving target detection and moving object tracking are focused on. Finally, a summary of the paper is concluded, and explaining the disadvantages and future plans. Following are details about the research contents and results of both moving target detection and moving object tracking.On moving object detection, this paper considers the advantages and disadvantages of optical flow method, frame difference method and background difference method. An improved moving object detection algorithm based on three-frame difference method and background difference method without background interference is proposed. Firstly, this algorithm detects moving targets through the combination of three-frame difference method and background difference method. Then, we detect and eliminate shadow of moving targets in HSV color space. Finally, the mathematical morphology is adopted to eliminate the effect of noise. Because of combining the advantages of frame difference method and background difference method, the proposed method can solve the problem of hollow inside the target and overcome the change of scene. In addition, it can eliminate the shadow and noise interference. The experiments show that the proposed algorithm can extract more complete outline of the moving objects and detect single target and multi-target without the shadow and noise interference. The results of indoor and outdoor moving object detection are good.On moving object tracking, some current main moving object tracking methods such as Mean-Shift, Particle Filter and TLD are introduced. The idea of Mean-Shift is to find the position of target through multiple iterations drift. In most situations, it could be easily realized to get accurate tracking position for low-speed targets. Mean-Shift algorithm is focused on, because it is a kind of fast and effective tracking algorithm. In addition, it has high performance of anti-interference. However, Mean-Shift algorithm doesn’t use the target’s motion direction and speed information in process of target tracking. So it brings about failures in fast motion target tracking. Therefore, the traditional Mean-Shift algorithm of the fixed tracking window cannot adapt to the size change of the target. An advanced Mean-Shift tracking algorithm using Forward-Backward Center Median and Three-Scale Information Measurement is proposed in this paper. Forward-Backward Center Median is used to adjust the position of the tracking window in each frame so that it can track fast motion object accurately. In the meantime, Three-Scale Information Measurement is used to change the size of the tracking window automatically. The experimental results show that the improved algorithm not only can help achieve fast and effective object tracking, but also can select the proper size of tracking window automatically to adapt to the increasing size or decreasing size of moving target.