Considering the problems that a large area of redundant coverage for heterogeneous sensor nodes in the initial network deployment, combining with relevant knowledge of geometry, and aiming at improving network coverage rate and the uniformity distribution of nodes, a coverage optimization algorithm based on reducing the area covered by redundancy between any two nodes through the nodes directional movement is proposed. The algorithm establishes decision threshold in advance. It makes the nodes implementation on directional shift by determining the magnitude of redundant coverage area with set threshold between any two nodes, and reduces the area of each overlap redundancy between nodes. Theoretical analysis and experimental simulation shows that the algorithm can effectively improve the deployment coverage of heterogeneous sensor network and optimize the distribution degree for nodes, especially with strong practicality for nodes deployment in large-scale sensor networks.