Abstract. In this paper, we conduct research on the SAR image target tracking algorithm based on the sparse representation and dictionary learning. The key technology of target tracking is to target can be extracted from the image detection, and to establish corresponding relationship with tracked target. Search target is extracted by testing on the whole image is very time consuming, in order to meet the requirements of real-time target tracking as should as far as possible to narrow the search scope in the entire image. By means of the criterion of correlation coefficient measurement to reflect the effect of target tracking, change according to the similarity of a certain strategy that is used to determine trace template, can effectively reduce the error updating the template. Under this background, our proposed methodology is feasible for enhancing the traditional approaches.