Research On A Class Of Adaptive Image Denoising Methods

According to related research,vision is the main way for human to acquire information.Images are the vivid description of objective things and the common information carrier in life.With the rapid development of science and technology,digital images were born.Due to their fast transmission and easy acquirement of information,they win the people's favorite.Digital image technology has greatly improved the speed of understanding of the world for people,while digital image noise is everywhere.Digital image noise causes interference to imaging,storage,and transmission processes.Image denoising is an important part of image preprocessing,which is the basis of deep image processing.In recent years,more and more noise appears with the increasing ways of digital acquisition.After using a lot of denoising methods,edge blur,noise residue and other issues will occur.Therefore,it is of great significance to study the image denoising.In this paper,three kinds of image denoising methods are improved.Firstly,for the wavelet denoising method,we divide the images into the smooth regions and the non-smooth regions according to the texture characteristic,and employ the improved new threshold function to respectively deal with the two parts.Secondly,for the compressed sensing denoising method,we use the correlation coefficient concept to improve the efficiency of the algorithm.Finally,for the block-matching and 3D filtering denoising method,we introduce the median filter as the preprocessing,and the variance estimation of the combined image is used to improve the related parameters in the basic estimation and final estimation.The improved methods have been applied to deal with various images as experiments.We first add the Guassian noise to the standard images,and compare before with after improvement of three methods.Subsequently,we make comparisons among improved methods by using new images.The experimental results show that the denoising effect is improved after improvement,and the improved methods are respecti,vely suitable for different kinds of noise.