A Proportional Algorithm for Rotary Motion Blurred Images Restoration
Yuchen Li, Fangxiu Jia, Xun Jiang, Guo Chuang · 2019
As for the rotation motion blur, the main reason for its poor recovery is that its blur kernel is difficult to estimate, which poses a new challenge to image restoration. In this paper, we propose a new blur kernel estimation model, which reconstructs the blurred image along the circumference concentric with the center of rotation. A series of blur kernels on the reconstructed image are easy to determine, and the blur kernel on each concentric circle is accurately described. A one-dimensional blur restoration technique such as Wiener filter can obtain the deblurred concentric circle image, and then use the inverse process of the model to refill the concentric circle image back to the Cartesian coordinate system to realize a high-definition rotating motion image restoration. The experimental results show that the proposed method can increase the Peak-Signal-Noise-Ratio (PSNR) of the rotational motion blurred image from 19.72dB to 31.42dB with less computational complexity.