首页 | 本学科首页   官方微博 | 高级检索  
     检索      

一种快速稀疏分解图像去噪新方法*
引用本文:李恒建,张家树,陈怀新.一种快速稀疏分解图像去噪新方法*[J].光子学报,2014,38(11):3009-3015.
作者姓名:李恒建  张家树  陈怀新
作者单位:(1 西南交通大学 信号与信息处理四川省重点实验室,成都 610031)
(2 中国电子科技集团公司第10研究所,成都 610036)
摘    要:提出了一种基于分层树型结构正交匹配追踪算法的快速图像去噪方法.通过选择高斯函数和墨西哥草帽小波母函数构建混合冗余字典,采用分层树状结构表示字典,结合构正交匹配追踪算法,实现图像稀疏表示,提高了图像表示的稀疏性,降低了算法的复杂度.依据噪音能量阈值,通过多次迭代达到图像去噪的目的.实验结果表明,在相同的噪音水平下,该迭代去噪算法取得了较高的较好的PSNR,获得更好的视觉效果.

关 键 词:图像处理  图像去噪  冗余字典  分层树型结构  正交匹配追踪
收稿时间:2008-04-29

A Fast Image Denoising Method Based on Space Decomposition
LI Heng-jian,ZHANG Jia-shu,CHEN Huai-xin.A Fast Image Denoising Method Based on Space Decomposition[J].Acta Photonica Sinica,2014,38(11):3009-3015.
Authors:LI Heng-jian  ZHANG Jia-shu  CHEN Huai-xin
Institution:(1 Sichuang Province Key Laboratory of Signal and Information Processing,
Southwest Jiaotong University,Chengdu 610031,China)
(2 No.10th Research Institute of China Electronics Technology Group Corporation,Chengdu 610036,China)
Abstract:A fast and efficient image denoising method is proposed based on Orthogonal matching pursuit which exploits the tree structure to reduce the complexity of the decomposition.Hierarchical clustering allows for building trees,where each node corresponds to molecule that encompasses the characteristics of all its relative children.Sparse decomposition can be fast solved by Tree based orthogonal matching pursuit,which leads to improved convergence.And,it guarantees the sparsity of results and exactly of reconstructed image.When denoising,the dictionary is built on the Gaussian function which suits well the task of capturing the low-frequency parts of nature image and the anisotropic function which has good ability to capture edges in images.The image denoising approach is taken based on mixture dictionary has the comparison performance with the state-of-the-art under the same noise level both visual quality and PSNR.
Keywords:Image processing  Image denoising  Redundant dictionaries  Tree structure  Orthogonal matching pursuit
点击此处可从《光子学报》浏览原始摘要信息
点击此处可从《光子学报》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号