A Non-uniform Motion Blur Parameter Identification and Restoration using Frequency and Cepstral Domain

A near accurate method for extracting blur parameters from a non-uniformly motion blurred images; in a blind image deconvolution scheme is proposed. In case of a non-uniform motion blur, we should be able to extract both the blur parameters and the combination of their extent fairly accurate, in order to improve the quality of the restored image. Initially, the parameters of the motion blur point spread function (PSF) of the observed blurry image are estimated. The blur parameters, which consist of two different directions and lengths of motion, can be extracted from the spectral and cepstral domain responses respectively, of that of the blurred image. Thereafter the morphological filtering is employed to enhance the precision of the directions and the lengths identification. Further, the estimated point spread functions (PSFs) of the motion blur are used to model the degradation function. A parametric Wiener filter performs deconvolution using the estimated PSF parameters and helps restoring these non-uniformly motion blurred images. The experimental results show that the performance of the algorithm proposed in this paper has higher PSF parameter estimation accuracy. General Terms Algorithms, Modeling, Performance, Estimation, Restoration.