Research Of Image Denoising Based On Blind Source Separation

Digital image is always affected by noise during digitaling or the transmission by the equipment and the environment factors.Image denoising has been a fundamental and indispensable research topic in the field of image processing,with the development of science and technology.Image denoising ideas and algorithms are more and more,and the denoising effect is better and better.Blind source separation plays a big role in the field of signal processing,and the so-called blind source separation is the process of getting the source data from observation data in the case of original data being unknown.Independent component analysis is an important way to solve the blind source separation,and it aims to find a method which can make the produced signals independent from each other.Now,the algorithms of blind signal processing are mostly based on independent component analysis.The basic idea of independent component analysis is to estimate the source signal through demixing observation signals,under the condition that the original signals are independent of each other.At present,the noise image is regarded as a mixed signal to be denoised in most of the denoising algorithms,so both in spatial domain and frequency domain,some denoising image signal algorithms have the disadvantages of denoising incomplete or abated.In this paper,the noise image is regarded as two independent signals,and then the noise image is separated into two signals which are image signal and noise signal by blind source separation method.The denoising effect is obtained obviously.For noise signal,this paper treats the high frequency noise information in the image as a virtual observation quantity by using the two-dimensional wavelet transform under the condition of independent component analysis mathematical model,and achieves the goal of denoising by using the method of blind source separation.Several traditional image denoising methods have been introduced in this paper,and some algorithms are discussed deeply.A new denoising method is given,and the method is also analyzed in the experiments.