Visual representation has an important influence on tracking performance.Unfortunately,most of the existing methods construct the visual representation only focus on the global target information, which ignored the information of the internal structure and the background of the target.To improve the accuracy of visual representation, a novel approach via local parts and context jointly is proposed in this paper.First, the target is decomposed into multiple independent patches based on the spatial structure, and each patch learns the filtering response separately.Second, the contextual information is introduced into the representation framework to migrate the background influences.With the introduction of collaborative representation strategy, the impact of background noise can be effectively suppressed.By conduct extensive experiments on several challenging benchmark datasets, the proposed approach demonstrates comparable performance against several state-of-the-art methods.