Fig. 1. Motion blur model projected by camera

Fig. 2. Process of generating camera motion trajectory

Fig. 3. The trajectory of the camera and the change of Euler angle under sinusoidal vibration excitation. (a) Trajectory of 3D space; (b) X-axis position over time; (c) X-axis Euler angle variation

Fig. 4. Sensitivity analysis of regularization parameters

Fig. 5. Blurred images and restoration results. (a) Blurred image; (b) algorithm in reference [7]; (c) algorithm in reference [16]; (d) algorithm in reference [19]; (e) proposed algorithm

Fig. 6. Aerial blurred images and restoration results. (a) Blurred image; (b) algorithm in reference [7]; (c) algorithm in reference [16]; (d) algorithm in reference [19]; (e) proposed algorithm

Fig. 7. Experimental plan flowchart

Fig. 8. Physical diagram of the experimental system

Fig. 9. Movement information of the UAV's onboard camera during exposure;(a)‒(c) Variation of the camera's triaxial Euler angle; (d) the trajectory of the camera in three-dimensional space; (e) projected trajectory on the image plane

Fig. 10. Real blurred images and restoration results. (a) Blurred image; (b) algorithm in reference [7]; (c) algorithm in reference [16]; (d) algorithm in reference [19]; (e) proposed algorithm

Fig. 11. Real blurred images and restoration results. (a) Blurred image; (b) algorithm in reference [7]; (c) algorithm in reference [16]; (d) algorithm in reference [19]; (e) proposed algorithm

Fig. 12. Iteration-varying of PSNR and SSIM. (a) PSNR of Fig.6; (b) SSIM of Fig.6; (c) PSNR of Fig.11; (d) SSIM of Fig.11

Fig. 13. Running time comparison of different algorithms. (a) Runing time of Fig.6; (b) runing time of Fig.11