Blind Separation for Permuted Alias Image Based Noise Detection

Permuted alias image is a new type of single channel blind separation,which is fundamentally different from traditional single channel blind separation in theory and method. In this paper, an algorithm about permuted alias images blind separation with noise detection is proposed according to a class of permuted alisa image infecting noise in permuting area. Firstly a mathematical model of permuted alias image is presented. Then a dictionary about sparse representation is obtained by training samples from permuted alias image with KSVD dictionary learning algorithm restrained by nonzeros number. The permuted alias image is denoised by utilizing learned dictionary. Size and location of permuting area is found out by detecting the subtract image, which is derived from result of the denoised permuted alias image subtracting original permuted alias image. The permuting area is optimized by the image morphological operation and is separated from the permuted alias image by making threshold. The results shows that permuting image can be separated efficiently from the permuted alias image, not affected by size, location, number of permuting image and noise level on permuting image.