Abstract: In the deep space exploration, astronomical optical autonomous navigation was used. The apparent magnitude of navigation stars for optical autonomous navigation was high. In order to obtain the clear star image, the star tracker's focal length and exposure time were long. Besides, star trackers were easily influenced by complicated motion. All these reasons resulted to the blurring and smearing star images. What was worse, it was difficult to restore the blurred star images by hardware methods. Studying the modeling method based on complicated motion was helpful to make sense the influence on the image quality which was caused by relative motion between the star tracker and the navigation star and to restore blurred star images by software methods. So that the star tracker's dynamic performance was improved. It was hard to simulate blurred star images which were caused by the complicated motion of star trackers for the lack of definite motion models. The angular motion and vibration were analyzed and an innovative method was proposed to simulate the blurred star images and then the ste-by-step simulation method was compared with the new method in this paper. The simulation results indicated that the simulation method based on the separable kernel could describe the real physical image process better than the ste-b-step method and the complicated motion had great influence on the extraction errors of star spot centroids, even leading to the failure of attitude determination of star trackers.