基于变换域的非局部均值图像去噪方法分析-analysis of non-local mean image denoising method based on transform domain.docxVIP

Abstract Title: NON-LOCAL MEANS IMAGE DENOISING METHOD BASED ON TRANSFORM DOMAIN RESEARCHMajor: Control Theory and Control EngineeringName: Li呼huai HUANG(Signatu陀:H h;tv1el严Supervisor:P时 Han LlU(Signature: 乒 -f仇守Lec. LiLi LlANG(Signatu阳Ii L ; LitJKl1)Abstract1n the process of collection ，transmission and acquisition，image is inevitably affected by noise interference. The interference reduces the image visual quali可. Therefore，before the image processing ，image de-noising is an indispensable pretreatment process. Non-local meansmethod is an effective de-noising method proposed in recent ye缸s. Through in-depth study of the non-local means ofthought ，this paper proposed two non-local means de-noising algorithms， based on dual-tree complex wavelet transform and singul町 value decomposition. The maincontents and conclusions 缸e as follows:Non-local means de-noising method based on dual-tree complex wavelet transformFor noise image in spatial domain is not sparse and non-local means weight 缸e susceptible to interference ，proposed based on non-local means de-noising method of dual-tree complex wavelet transform. The method makes full use of the DT-CWT has translational invariance and directional selectivity，to solve the noise image in spatial domain is not easy to sparse and non-local means weight calculation inaccuracies. Compared to the original non-local means and its major improvements de-noising methods ，the proposed method has a certain competitive inthe de-noising effect.Non-local means de-noising method based on singular value decomposition1n order to further improve the image de-noising effect，proposed non-local means de-noising method based on singular value decomposition. Firstly，the method used the non-local means，similar image patches 缸e simply collected together from a noisy image with block matching. We directly use these 2-D patches to calculate their 2-D SVD basis andtransform them into 2-D SVD domain. Since the 2-D SVD can prese凹e image nature structur