Font Size: a A A

Research On Signal Recoveryalgorithms Based On Compressed Sensing

Posted on:2013-01-02Degree:MasterType:Thesis
Country:ChinaCandidate:W Z JiFull Text:PDF
GTID:2218330368488732Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
In recent years, with the rapid development of modern information technology, conflicts between the demands for information and the traditional signal processing theory become more and more serious. Nyquist sampling theorem must be satisfied for accurately recovering original signal, but in practical applications, a high sampling rate will produce large amounts of raw sampling data, and substantially increase information transmission, storage and processing pressure. In this context, Candès and other scholars found a new signal processing method -- Compressed Sensing. For sparse or compressible signal, compressed sensing theory uses frequency which is far less than the sampling rate that Nyquist sampling theorem required, sampling and compressed are implemented successfully, and can accurately recover the original signal.Signal recovery algorithms is the core of compressed sensing, the key question to study signal recovery algorithms is how to use the low-dimensional measured signal to accurately recover the original high-dimensional signal. Therefore, this thesis investigates signal recovery algorithms based on compressed sensing. The whole thesis is consisting of four parts.Firstly, briefly introduces compressed sensing theoretical framework, detailedly analyzes sparse representation of signal, linear measurement and signal recovery, describes the initial applications of compressed sensing theory, which lays a theoretical foundation for further study.Secondly, concludes some common signal recovery algorithms, then focuses on gradient projection for sparse reconstruction and compressive sampling matching pursuit algorithm, verifies the performance of the algorithms, and finally improves compressive sample matching pursuit algorithm, proposes a new atom update method, results show that improved compressive sampling matching pursuit algorithm is superior to existing compressive sampling matching pursuit algorithm for the image signal recovery quality under the same conditions.Then, introduces orthogonal matching pursuit algorithm, for the disadvantage of the algorithm performance, uses atom selection ideas of optimized orthogonal matching pursuit algorithm, this thesis improves orthogonal matching pursuit algorithm. By Matlab simulation platform, carries out signal recovery experiments with one-dimensional discrete signal and two-dimensional image, results show that improved orthogonal matching pursuit algorithm has better signal recovery quality and shorter running time under the same conditions.Finally, concludes the whole thesis and presents some future work.
Keywords/Search Tags:Compressed Sensing, Sparse Presentation, Matching Pursuit, Signal Recovery Algorithms, Restricted Isomety Property
PDF Full Text Request
Related items