An Improved Blind Image Restoration Based on Bi-Spectral Reconstruction

Abstract

Image blind restoration is very important in our life. The image restoration is a ill-posed question so the regularization is much better method. For the regularization method, the most important is to select the regularization parameter [1]. If the parameter is bigger, to be smooth the edge or detail, but smaller, not to be smooth the noise [2], In this paper, we present a new method. Firstly, decomposing the image using wavelet transform, the high frequency information is corresponding to the edge and noise, the low frequency is the flat .We denoise using the bi-spectral reconstruction in high frequency, for the low frequency, we recover by the regularization method .This method has advantage in holding the edge and is simple to choose the parameter of regularization .Experimental results show the good performance, this method is very effective for the image polluted by the symmetry noise.