电子学报 ›› 2017, Vol. 45 ›› Issue (3): 695-703.DOI: 10.3969/j.issn.0372-2112.2017.03.029

• 学术论文 • 上一篇    下一篇

基于非局部相似块低秩的压缩感知图像重建算法

宋云1,2,3, 李雪玉1,2,4, 沈燕飞4, 杨高波3   

  1. 1. 长沙理工大学综合交通运输大数据智能处理湖南省重点实验室, 湖南长沙 410114;
    2. 长沙理工大学计算机与通信工程学院, 湖南长沙 410114;
    3. 湖南大学信息科学与工程学院, 湖南长沙 410012;
    4. 中国科学院计算技术研究所, 北京 100190
  • 收稿日期:2015-08-05 修回日期:2015-11-10 出版日期:2017-03-25
    • 作者简介:
    • 宋云 男,1974年9月出生于湖南省汨罗市.长沙理工大学计算机与通信工程学院副教授,硕士研究生导师,现于湖南大学信息科学与工程学院攻读博士学位.主要研究方向包括数字图像处理、视频编解码、压缩感知和计算机视觉等.E-mail:sonie@126.com;李雪玉 女,1991年1月出生于河南省信阳市,长沙理工大学硕士研究生,中国科学院计算技术研究所客座学生.主要研究方向为数字图像处理、视频编解码和计算机视觉等.E-mail:lixueyu_17@163.com;沈燕飞 男,1976年4月出生于江苏省靖江市,中国科学院计算技术研究所副研究员,博士.主要从事数字图像处理、多媒体通信和计算机视觉等方面的研究工作.E-mail:syf@ict.ac.cn;杨高波 男,1974年7月出生于湖南省岳阳市,湖南大学信息科学与工程学院教授,博士研究生导师,主要研究方向为数字图像处理、视频编解码、数字图像安全等.E-mail:yanggaobo@hnu.edu.cn
    • 基金资助:
    • 国家自然科学基金 (No.61471343,No.61572183,No.61402053); 湖南省教育厅科学研究重点项目 (No.13A107,No.15A007); 湖南省自然科学基金 (No.2016JJ2005); 湖南省科技计划项目 (No.2014FJ6047,No.2014GK3030)

Compressed Sensing Image Reconstruction Based on Low Rank of Non-local Similar Patches

SONG Yun1,2,3, LI Xue-yu1,2,4, SHEN Yan-fei4, YANG Gao-bo3   

  1. 1. Hunan Provincial Key Laboratory of Intelligent Processing of Big Data on Transportation, Changsha University of Science and Technology, Changsha, Hunan 410114 China;
    2. School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha, Hunan 410114 China;
    3. College of Computer Science and Electronic Engineering, Hunan University, Changsha, Hunan 410012 China;
    4. Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China
  • Received:2015-08-05 Revised:2015-11-10 Online:2017-03-25 Published:2017-03-25
    • Supported by:
    • National Natural Science Foundation of China (No.61471343, No.61572183, No.61402053); Key Program of Science Research of Education Department of Hunan Province (No.13A107, No.15A007); Natural Science Foundation of Hunan Province,  China (No.2016JJ2005); Science and Technology Project of Hunan Province (No.2014FJ6047, No.2014GK3030)

摘要:

传统的压缩感知重建算法利用信号在某个特征空间下的稀疏性构建目标优化函数,但没有充分考虑信号的局部特性和结构化属性,影响了算法的重建性能和算法的适应性.本文考虑图像的非局部自相似性(Nonlocal Self-Similarity,NLSS),提出一种基于图像相似块低秩的压缩感知图像重建算法,将图像恢复问题转化为聚合的相似块矩阵秩最小问题.算法以最小压缩感知重建误差为约束构建优化模型,并采用加权核范数最小化算法(Weighed Nuclear Norm Minimization,WNNM)求解低秩优化问题,很好地挖掘了图像自身的信息和结构化稀疏特征,保护了图像的结构和纹理细节.多个测试图像、不同采样率下的实验证明了算法的有效性,特别是在低采率下对于纹理较为丰富的图像,提出的算法图像重建质量较明显的优于最新的同类算法.

关键词: 压缩感知, 图像重建, 非局部自相似, 低秩优化

Abstract:

Generally,traditional compressed sensing (CS) image recovery methods build the objective optimization function by using the signal sparsity in some specific feature spaces.They do not fully take the local features and structural properties of signal into account,which leads to constraints of the recovery performance and flexibility.In this paper,considering the non-local self-similarity (NLSS) in images,we propose an image CS reconstruction method based on the image low-rank property by converting the CS recovery problem into a matrix rank minimization problem of aggregating similar image patches.The proposed algorithm builds optimization model under the constraint of minimal recovery errors and employs the weighed nuclear norm minimization (WNNM) method to solve the low-rank optimization problem.By taking advantage of them,the proposed method exploits the self-information and structured sparse characteristics of the image very well,and therefore provides a better protection of image structures and textures.Experiments on different test images under various sampling rates have shown the effectiveness of the proposed algorithm.Especially,for richly-textured images,our method outperforms the art-of-the-state algorithms significantly under low sampling rates.

Key words: compressive sensing, image recovery, non-local self-similarity, low-rank optimization

中图分类号: