Motion Blur Parameter Identification from a Linearly Blurred Image

Restoration of motion blurred image is a highly demanded technique. The motion blur parameter identification plays an important role in the image restoration. In this paper, we propose the blur parameter identification algorithm under linear motion blur. The parameters to be estimated are the length and direction. The Fourier spectrum of the image degraded by linear motion blur has regular patterns of zeros. The proposed algorithm detects those patterns by minimizing the correlation between the Fourier spectrum of blurred image and a detecting function. The experimental comparisons show that our proposed algorithm has robustness against noise than other two algorithms.