Real-time restoration of images blurred by rotational motion

A real-time image restoration algorithm is proposed for the images blurred by rotational motion during aerial imaging. According to the rotational motion with space-variant blur and different motion rates for image, this paper analyzes how the rotational motion is generated and derives a mathematical model of rotational motion. Then, the coordinate conversion based on Bresenham algorithm is applied to transform the space-variant blur on circular arc into the space-invariant blur on pixel lines and 1D Wiener filter is used to restore the linear image motion transformed by coordinate conversion. The experimental results show that the proposed algorithm can restore a 10241024, 8 bit blurred image in 3.31 ms on the Graphics Processing Unit (GPU) platform, and the details in the rotational motion blurred image is pretty well restored and the PSNR of the restoration image can reach 28.76. The restoration quality and the effectiveness of the proposed algorithm reveals it is pratical in the aerial imaging system.