Motion blur parameters estimation based on frequency and spatial domain analysis
Yu Shen, Jianwu Dang, Tao Lei, Weiwei Luo · 2012
Restoration of motion blurred image is highly dependent on motion blur parameters estimation, such as motion length (L) and motion direction (Φ). After studying the nature of the degradation function and the point spread function (PSF), the paper presents an improved algorithm to estimate linear motion blur parameters. The algorithm is based on the combination of frequency and spatial domain analysis. In this algorithm, Radon transform is used to find the motion direction in the frequency domain, and autocorrelation method is used to estimate the motion length in the spatial domain. The experiment results show that this algorithm can effectively estimate motion blur parameters.