A stellar image denoising method for active pixel sensor (APS) based on dynamic sequential noise template is proposed. In this method, a dynamic sequential noise template is constructed based on the short-term static background noise and the random noise threshold, which is used to realize the denoising treatment of APS stellar images. The original APS stellar image of the ZY-3 is used as the test data and the research results show that the proposed method can eliminate strip and edge noises very well. Compared with the traditional denoising method, the proposed method can improve the extraction accuracy of the stellar image centroid by approximately 12.98″. Compared with the global threshold segmentation method, the proposed method can improve the extraction accuracy by about 2.16″. Compared with the static background noise method, the proposed method can improve the extraction accuracy by around 9.64″. These fully indicate that the proposed method can effectively improve the positioning accuracy of APS stellar centroid and can be used in attitude post-processing.