Study Of Tracking Algorithms Of Moving Targets On Indoor Environment

Moving target tracking as one of the hot issues in video monitoring, has been take care of by many experts and scholars. A lot of tracking method have been put forward at present.Moving target detection is the premise of the tracking. Through the detection if there are moving targets and the number of the moving targets in the video image sequence can be judged. According to the detected target information, use the target characteristics correlation can make the targets tracking respectively. So as to get each moving object location information. High precisioning,strong robustness and good real-time is the target tracking requirement. A variety of moving target detection and tracking method are researched in this paper. In order to find the optimal solution on the basis of each method, This paper mainly makes the human body target as the research object. Under the condition of the camera on the tripod be fixed, to detect and track the moving targets in the indoor environment (static background). Mainly study the three aspects that the motion target detection, target feature extraction and matching and moving target tracking.1. Moving target detection, through the analysis of optical flow method, background subtraction and phase frame differential method principle and their respective advantages and disadvantages, puts forward a multiple moving target detection method that make the fusion of the three frame difference and background subtraction. The experiment shows that the method can extract target outline effectively, detect the moving target for real time.2. In the course of analysis of moving target features (color, edge, texture, angular point, etc.), mainly research color feature extraction based on the color histogram, texture feature extraction based on gray level co-occurrence matrix, angle point feature extraction based on harris corner detection and matching and the edge feature extraction method based on the gradient operator. Feature extraction and matching method provide a premise for moving target tracking accuracy.3. Target tracking. Because of the lights in the environment changes, target and background color similarity of issues. The particle filter method based on the second order histogram, that is, using the color feature matching method is putted forward. The combination of improved Camshift (Continuously Adaptive Meanshift) method to track moving targets based on color, texture characteristics is raised too. The experimental results show that tracking effectiveness and correctness which use the proposed improved particle filter method and improve Camshift algorithm is better than the Meanshift method based on color feature, and can effectively overcome the interference problems that because of the similarity of the color. However, the above two methods can’t satisfy the occlusion of target tracking accuracy, fast, and robustness requirements, the paper will make the two methods combined effectively, if there are no barriers the improved camshift algorithm will be used. Once the blocks occur, particle filtering algorithm is used immediately,when. This combination of algorithm mainly uses the Camshift algorithm faster and predictive particle filter more accurate, so as to achieve tracking requirements.