Research On Object Tracking Algorithm Based On Sparse Representation

In recent years,the target tracking technology has been widely used in many fields.However,in practical application,as the tracking algorithm is affected by occlusion,illumination changes,rotation and translation,complex background and other factors,the real-time and accuracy of tracking is sometimes difficult to meet the actual needs.In the tracking algorithm,particle filter algorithm can effectively fit the motion state of the target,and sparse representation can highlight the main features and key characteristics in the image.Therefore,in the framework of particle filter,the sparse representation theory is introduced to the target tracking,we propose a target tracking algorithm based on sparse representation.The main research work of this thesis is as follows:1.Aiming at the problem of large matching error between the target template and the candidate template in the global representation when the target is blocked,a target tracking method based on local sparse representation is been proposed.In sparse representation of the target,the target region was divided overlapped,and each local image block was normalized,the normalized local image blocks were sparsely encoded,and then we can obtain the local information of the target,thus the target anti-blocking ability is improved.In the establishment of the target function,local image blocks were processed with sparse coefficient weighted preconditioning,and the similarity function was designed to calculate the similarity between the target and the target template,which can further reduce the influence of the occlusion factor,so as to obtain the best estimation of the target position.Simulation results show that compared with the classical target tracking algorithm,this algorithm has better performance when the target is occluded,and to a certain extent,it can overcome the ineffective performance of the global template in dealing with the local change of the target,and improve the accuracy and robustness of the algorithm.2.Aiming at the problem of tracking drift and low efficiency caused by partial occlusion,rotation and scale change in the target tracking,a new target tracking algorithm based on color histogram and local sparse representation is proposed,which is based on the particle filter framework and sparse representation.In the construction of the target template,the first we used the global and local templatesto build collaborative template,and then calculated the similarity between the candidate target and the target template by designing metric function,thus,the position of the target in the current frame can be determined.In the process of tracking,we defined two target observation models.Template had strong discriminative and adaptability by occlusion judgment,the algorithm can adaptive dynamic select observation model and adopt the optimal template update strategy.Experimental results show that the proposed method has better robustness in the case of target rotation,partial occlusion and scale variation.