压缩感知图像重构中矩阵互相关性的研究.pdfVIP

Image Pr∞essing and Multimedia Technology 压缩感知图像重构中矩阵互相关性的研究 郑永奇 1 ，焦铸 2 ，韩玉霞 1 (1.郑州谢青医学高等专科学校，河南郑州 453400 j 2. 西安理工大学，陕西西安 710068 ) 摘 要:从测量矩阵和稀疏矩阵的王相关性角度出友，通过对测量矩阵和稀疏矩阵所构成的 Gram 矩阵进行门限选择，进而经过相应的缩放处理降低互相关性，这样不仅可以获取更多有信息量 的测量值，而且可以完成对测量矩阵的优化改进。通过在 DWT 、 DCT 下的压缩感知图像重构实验验证 了该方法的可行性，恢复效果得到一定程度的提高，相比于传统的小波恢复重构，达到了预期的效果。 关键词:压缩感知;稀疏表示;测量矩阵 中圈分类号: TN919 文献标识码 :A 文章编号: 1674-7720 (2013 )05-0042-04 Study of correlation between matrix in compressed sensing image reconstruction Zheng Yongqi 1 .Jiao Zhu2 • Han Yuxia 1 (I.Zhengzhou Shuqing Medical College , Zhengzhou 453400 , China; 2.Xi an University of Technology , Xi an 710068 , China) Abstract: This paper begins with the perspective of the correlation between measurement matrix and sparse matrix. By making threshold selections of Gram matrix constituted by measurement matrix and sparse matrix , and making appropriate scaling processmg , the correlation is reduced. It can not only obtain more measured value with information content , but also finish the work to improve the measurement matrix. Through the DWT and DCT compressed sensing image 陀construction experiments , the feasibility of the method is verified and the recoverγresults have been improvcd to some extent. Compared with the conventional wavelet restoratioll recollstruction. it achieves the desired results. Key words: compressed sensing; sparse representation; measurement matrix 传统理论指导下的信号处理首先遵循奈奎斯特/香 研究分析的基础上进行了优化改进，并通过实验模拟 农采样定理对信号进行采样，再对采样数据进行某种 进行了有效的验证。 变化(如 DCT 或 DWT) ，将变化后幅值较大的系数进行 1 压缩感知的基本理论 压缩编码，舍弃幅值为零或接近零的系数，然后将编码 压缩感知发展于信息论，由于其具有高度信息挖掘 值