Research On Object Tracking Algorithm Based On Sparse Representation

Object tracking is an important content in the field of computer vision, which is a hot topic in recent years. It has many important applications such as video surveillance systems, intelligent traffic monitoring, human-computer interaction and military. With the problems of object tracking, this paper using visual characteristics of sparse representation, introduced the sparse representation theory into the particle filter framework to achieve the target tracking in different situations such as occlusion, illumination changes, object moving fast and similar backgrounds.The main work of this thesis is as follows:(1) On the basis of analyzing and summarizing the research status at home and abroad of object tracking and key technology, and achieved a traditional target tracking method based on sparse representation, through the experiment, analysis the advantages and limitations of this tracking method;(2) Due to the traditional object tracking method based on sparse representation cannot well tracking in the scene with object moving fast and the background similar to the tracking object, we implemented an object tracking algorithm based on sparse representation joint template, the object appearance model can sparse representation simultaneously with the original object template and the template built by some particles in the current frame, and we build a joint objective function which will solve the tracking issue through an iterative optimization problem, by the particles with sparse coefficient to complete object tracking, and achieved better tracking results;(3) An object tracking with local sparse representation is researched and implemented, we obtained the sparse coding of corresponding local image block of the object area, with all of the sparse coding of the whole object to represent the object, and learned an online classifier to distinguish the target from the background, and finally tracking with the online classifier in a two-stage tracking algorithm.