The Research On Particle Filter Tracking Based On Multi Feature Fusion Algorithm

Object tracking is an evolving research field in the field of computer vision,which incorporates many fields related technologies such as image processing,pattern recognition,artificial intelligence and machine vision.Due to the continuous research in the field of target tracking,it is widely used in many fields such as urban security,intelligent transportation,satellite launch,human-computer interaction and so on.Therefore,many excellent scholars involved in the target tracking study,put forward many excellent tracking algorithm.The particle filter algorithm is widely used in target tracking because of its good characteristics in dealing with sublinearity and non-Gaussian system.The traditional particle filter algorithm only uses a single feature to describe the target,and it is difficult to track the target quickly and accurately under the influence of interference such as occlusion,self-deformation and illumination change.Due to the long tracking,the error accumulates,resulting in the tracking window Offset,the ultimate goal is lost.The multifeature fusion method is usually used to improve the tracking performance of particle filter.The main work of this paper are list as follows:A particle filter tracking algorithm for single feature is used to track the problem of poor target in complex scenes.This paper presents a multi feature fusion particle filter target tracking algorithm.The color feature and texture feature are used to describe the target.The color feature is not sensitive to the rotation of the target and the change of the scale.The treatment of the occlusion problem is better,but it is easy to be affected by the light.The texture features are not sensitive to light,The feature can be complementary in the tracking process,and the fusion coefficient of the feature can be adjusted adaptively by the characteristic uncertainty factor.And the mean shift algorithm is incorporated into the particle filter frame.Through its convergence,the sample particles are converged as far as possible to contain the target real position.At the same time,in the process of tracking,according to the accuracy of the target tracking changes in the number of adaptive update particle set,to enhance the adaptability of the algorithm on the environment.The experimental results show that the proposed algorithm can track the target accurately and has good robustness and real-time performance when it is in the complex scene and occluded.