Fig. 1. Illustration of satellite self-rotation and precession

Fig. 2. Flow chart of satellite motion parameter identification algorithm

Fig. 3. Diagram of circle fitting

Fig. 4. Diagram of relationship among motion parameters

Fig. 5. Function introduction of simulation software. (a) Size of satellite model and initial azimuth information of sensor model;(b)-(d) parameters of camera model, setting interface of motion parameters, and data acquisition interface in simulation software

Fig. 6. Illustration of satellite data. (a) Color image; (b) depth image; (c) 3D point cloud

Fig. 7. Registration error between two adjacent point clouds. (a) α; (b) β; (c) γ; (d) tx; (e) ty; (f) tz

Fig. 8. Comparison between two adjacent point clouds before and after registration. (a) Before registration; (b) after registration

Fig. 9. Results of circle fitting under different Gaussian noise. (a) σ=0.01 m; (b) σ=0.02 m; (c) σ=0.03 m; (d) σ=0.04 m; (e) σ=0.05 m; (f) σ=0.06 m; (g) σ=0.07 m; (h) σ=0.08 m; (i) σ=0.09 m

Fig. 10. Errors of center of circle, radius and normal angle of fitting circle under different level of noise.(a) Center of circle; (b) radius; (c) normal vector

Fig. 11. Estimated error of each group of three estimated parameters. (a) Angular velocity of precession; (b) angular velocity of spin; (c) angle of nutation

Fig. 12. Parameter identification error changes with point cloud noise intensity. (a) Angular velocity of precession; (b) angular velocity of spin; (c) angle of nutation

Fig. 13. Illustration of satellite data. (a) Color image; (b) depth image; (c) 3D point cloud

Fig. 14. Intermediate results of parameter identification. (a) Point cloud of satellite; (b) motion track of point on satellite; (c) distribution of estimated speed of self-rotation