Aimed at the problems of partial occlusion, background interference and other issues in visual object tracking technology, the concept of superpixel is proposed. Firstly, Simple Linear Iterative Clustering (SLIC) method is used to segment the target and the surrounding area into superpixels based on texture feature represented by local entropy and color feature. Then confidence map is established by the appearance of training model to endow each pixel with the confidence value. Updating the appearance model ensures the tracking process, which can adapt to the changes of target and background. Finally, the location of the target is found according to the confidence map. The experimental results show that the proposed tracking algorithm can effectively find the target position in the case of partial occlusion or background interference.