基于方法噪声稀疏表字典学习的图像去噪算法  

Image Denoising Algorithm Based on Method Noise Sparse Representation Dictionary Learning

作　　者：黄丽韶[1] 文海英[1] 顾思思[1] 

机构地区：[1]湖南科技学院电子与信息工程学院,湖南永州425199

出　　处：《系统仿真学报》2016年第1期154-159,166,共7页Journal of System Simulation

基　　金：湖南省教育厅科学研究项目(13C336)

摘　　要：针对图像去噪过程中会导致部分纹理信息丢失的不足,提出了基于方法噪声稀疏表示的图像去噪算法。该算法基本思想是通过对方法噪声(受噪声污染的图像和去噪后图像的差)的稀疏表示,提取方法噪声中的纹理信息,从而提升图像去噪质量。采用导引滤波去噪得到图像的方法噪声;利用该方法噪声通过改进的字典学习算法训练得到自适应的冗余字典;利用该字典提取方法噪声图像中的纹理信息,由导引滤波去噪后的图像和提取的纹理信息恢复图像,达到去噪效果。实验结果表明,所提出的方法峰值信噪比高于现有的同类算法,且能较好地保持图像的细节和纹理信息,提高了视觉效果。For the shortcoming of losting partial texture information with image denoising process, the image denoising algorithm based on method noise sparse representation was proposed. The method noise, which was defined as the difference between the noisy and the denoised image, was obtained by guided filter. Then redundant dictionary was learned by improved dictionary learning method and the method noise. The image texture information in method noise was extracted by the learning dictionary, and image was restored by the extracted image texture information and denoised image by guided filter. The experimental results demonstrate that the peak signal to noise ratio Value（PSNR） of the proposed algorithm is better than state-of-the-art algorithms, while the proposed algorithm can well preserve the texture information in the denoised image, making them look more natural.

关 键 词：方法噪声 字典学习 冗余字典 稀疏表示 

分 类 号：P391.41[天文地球—地球物理学]