To solve the occlusion problem of multiple moving objects in video surveillance, this paper proposes a novol target tracking algorithm based on combination of global feature matching and local feature matching. In this algorithm, the methods based on histogram and block segmentation are adopted to express gray features of the target. The occlusion is pre-judged before it happens. When there is occlusion, the gray features based on block segmentation are used to track target. After occlusion, the target is re-located by histgram matching. The experiment results demonstrate the validity and superiority of the proposed approach.