Research On Noise Reduction Based On Image Sequence

Today's information age,we always touch the image.Interference of various factors will lead to the destruction of information on the image,affecting people's access to information and follow-up treatment.Based on Blind Source Separation(BBS)and the separation theory of Independent Component Analysis(ICA),The essence is that the noise and image data are independent of each other and it can used to reduce noise.In this paper,ICA theory is applied to image denoising,based on ICA combined with sparse coding shrinkage method for real image sequence and simulated image sequence noise reduction,Algorithm complexity by changeing input image sequence number.Then,an evaluation criterion HMSE based on high-order non-reference is proposed.In the process of experiment simulation,it is found that the evaluation standard can be used to evaluate the noise reduction effect objectively.This paper is based on MATLAB2015 b,FastICA algorithm is compared with Frame Average algorithm and Singular Value Decomposition(SVD)algorithm.The noise reduction characteristics are studied by changing the sampling number and noise standard deviation.The results show that the performance of multiple image sequence noise reduction algorithms is superior to other single image denoising methods.In general,this practical and promising X-ray image denoising method does not need the prior information of the noise,it can extract the useful image information,the extraction efficiency increases with the number of images in the image sequence,and will also improve with the technical improvement of the blind source separation method.