The Research For Target Tracking Algorithms Based On Particle Filter

Target tracking in video sequence is an important branch of computer vision. The theory of target tracking combines advanced technologies and research achievements in image processing,pattern recognition,artificial intelligence,automatic control and relative fields. Target tracking algorithm is based on the classic Gauss Filtering Theory. At non-linear and non-Gaussian status for moving target, the estimation cannot be solved by traditional linear filter. Recursive Bayesian Filter can be realized by Mont-Carol simulation for particle filter. It can be applied for any non-linear and non-Gaussian system indicated by status space module and precision approach to the best estimation. The topic of this paper has promising prospect and competence. It can satisfy with research, manufacture, military use, and so on.Noise in the image is analyzed and pre-processed. The theory and implementation of particle filter is introduced, which is used to solve the problem of target tracking. The framework of target tracking algorithm based on particle filter is proposed. With the gray-feature of target, correlation-based tracking based on particle filter is studied. The simulation result shows that the proposed algorithm has good performance. Also the accuracy of the algorithm is quantitative analyzed. Considering particle number and particle radius, which are adjustable, correlation-based tracking based on particle filter has the merit of high accuracy and robust anti-jamming ability compared with the traditional algorithm.