基于自适应低秩去噪的近似消息传递压缩感知恢复
Compressive sensing reconstruction based on approximate message passing with adaptive low-rank denoising
投稿时间:2018-07-19  修订日期:2018-07-19
DOI:
中文关键词: 压缩感知恢复  近似消息传递  低秩去噪  迭代阈值
英文关键词: Compressed Sensing(CS)  Approximate Message Passing(AMP)  low-rank denoising  iterative threshold
基金项目:国家自然科学基金资助项目(61471400)
作者单位E-mail
熊承义 中南民族大学 xiongcy@mai.scuec.edu.cn 
陈仕长 中南民族大学  
高志荣 中南民族大学  
李世宇 中南民族大学  
金鑫 中南民族大学  
李治邦 中南民族大学  
摘要点击次数: 244
全文下载次数: 
中文摘要:
      图像隐含的低秩先验特性已被成功应用于去噪等图像恢复应用.考虑到自然图像具有的非平稳特性以及迭代重构中图像噪声强度的变化,提出了一种结合近似消息传递与自适应低秩去噪的图像压缩感知重构方法.根据迭代重构图像的噪声方差估计,自适应地调整分块图像的大小以及相似块组的规模,实现低秩去噪性能的有效提升,从而保证了迭代重构的收敛速度,并同时改善了重构图像的质量.大量实验结果表明:该方法在无噪和有噪观测环境下都具有较好的重构性能,且能够有效的保留图像的纹理细节信息.
英文摘要:
      The implicit low-rank priori characteristics of images have been successfully applied to image restoration applications,such as denoising. Considering the non-stationary property of natural images and the variation of image noise intensity in iterative reconstruction, an image compressive sensing reconstruction method based on approximate message passing and adaptive low-rank denoising is proposed.According to the noise variance estimation of the iteratively reconstructed image, the size of the block image and the size of the similar block group are adjusted adaptively to achieve an effective improvement of the low-rank denoising performance, thereby ensuring the convergence speed of the iterative reconstruction and simultaneously improving the quality of the reconstructed image. The experimental results show that the proposed method has good reconstruction performance in both noiseless and noisy observation environments, and can effectively preserve the texture details of the image.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭