压缩感知及应用.pdfVIP

微计算机应用 Vol. 31 No.3 第 31 卷第3 期 Mar.20 1O 2010 年3 月 MICROCOMPUTE日 APPUCATIONS 压缩感知及应用* 李卓凡1 ， 2 闰敬文2 e. 韩山师班学院物现与电子工程系 潮州 521041 1 灿头大学工学院汕头 515063 ) 销襄:传统的信号采样必须谦循香农采样定理，产生的大盘数据造成了存储~间的浪费。压缩感知(CS) 提出一种新的来样理 论，它能够以远低于奈奎斯特来样遮盖棋来样倍号。压缩J感知的慕本论点是如果信号具有稀疏性，可投影到一个与变换基不相 关的随机矩阵并获得娼少子信号长度的祖tl 囊僚，再通过求解优化问题，精确道构倍号。本文详述了压缩感知的基本理论，压 缩感知适用的基本条件:稀疏性和非相干性，测囊矩阵设计要求，及重构算法的阳P 准则，并介绍了压缩感知的应用及仿真。 仿真结果表明当果样个数大于 K xlog(N/旬，就能将 N 维信号稳定地重建出来。 关键词:压缩感细观测矩阵稀疏性 RIP Theory and Applicaton of Compressive Sensing U Zhuofan ,YAN Jingwen ( I Physics and electronic engineering department ，日部lShan Normal College , Chaozhou ,521041 , China; 1College of Engineeri吨，Shantou University ,Shantou , Guangdong ,515063 , China) Abstract: Conventional approaches to sampling signals follow Shannon principle. It take great costs on data storage. In this pa阳r ， the theo巧。f Compressive sensing is introduced. Compressive 朋nsing provides a new sampling thoory to sample signal below the Nyquist rate. If si伊aI or image is s阳rse in some orthonormal basis , si阴aI or image c栩如何covered from small number of measurement using an optimization proceS8 . The structure of 由e si伊aI is preserved in the measurement and the measur暗 matrlx is incoherent with the or- thonormal basis. CS relies on two principles: 印刷ity and incoheren伺. RI pprinciple is the precondiction of desi伊i吨 reconstru悦。n algorithm. The application of CS thooηare introduced and the simulation is illustrated in details. The simulation show that the sig