Deformable Model Based Human Motion Tracking Analysis

With the development of science and technology, human motion tracking has developed into a multi-disciplinary overlapping research area. As the direction of advanced research field in image processing and computer vision, it has a wide application prospect and commercial value in many fields.More and more research institutions and business organizations have been put into the field of motion tracking. This paper focuses on the technology of human motion tracking based on model. First,we use the improved color space conversion technology to extract the moving object; then statementing the deformable model; at last using four kinds of algorithm to track human motion and analysis the advantages and disadvantages of them.The main work this paper does is as the following:(1) Using color space conversion method to detect and segment moving objects. In view of the actual situation in this experimental, the traditional color space conversion algorithm is improved and using moving targets’good clustering in the improved YCrCb color space to achieve the segmentation of moving objects.(2) Establishing human model. This paper uses a deformable model to model the limbs of human body, the model can approximate reflect the arm and leg muscles deformation during motion tracking process.(3) Motion tracking based on model and quality analysis. Under the same kind of deformable model, we realized particle filter, template matching, discriminant and generative algorithm, covariance scaled sampling four kinds of tracking algorithm to track human moving respectively and analysed tracking results from multi-sapects like robustness, accuracy, real-time to summarize the advantage and defects of each.This paper use two videos which we shoot ourselves as the experiment material, using four kinds of algorithm described above to realize human motion tracking. By the experimental results, analysing four algorithms’advantages and disadvantages in robustness, accuracy, real-time and other aspects and giving the guidance.