With the development of remote sensing, the data quantity of remote sensing image is increasing tremendously. It brings a huge workload to the image geometric correction through manual ground control point (GCP) selection. GCPs automatically selected based on software is one of the effective methods to cut down manual operation. The GCPs obtained from that way is generally redundant. This paper deeply comprehends some existing methods about automatic optimization of GCP, and puts forward a new method of automatic optimization of GCP based on Thiessen Polygon to filter ground control points from the overfull ones without manual subjectivity for better accuracy. Experiments in this paper also demonstrated the relationship between the accuracy of geometric correction and the distribution of GCP. It advances the conception of single GCP’s importance value based on Thiessen Polygon. The paper gives the theory and the flow of automatic optimization of GCPs as well. It also presents an example of the application of this method. In the conclusion, this paper points out the advantages of this method .