| The essential attributes of traditional Nyquist sampling theorem consume a lot of storageresources and bandwidth resources. And the compressive sensing theorem focuses the signalsampling and data compression of the process of sampling theorem into one step, only to considerthe characteristics of the signal structure, instead of considering the limitation of signal bandwidth.Thereby it can significantly reduce the requirements of sampling bandwidth and save storage spacefor data. According to the advantage of compressed sensing, it has been applied to more and morefields in real life, mainly contain the fields of wireless communications, image processing,geographic information, radar imaging, pattern recognition and so on.Compressed sensing is put forth by the extensive attention after appearing, researchers mainlyexplore three key directions of compressed sensing: sparse signal expression, measurement matrixdesign and reconstruction algorithm construction, and have achieved large number of breakthroughsand progresses. The first two of directions are prerequisite conditions for usage of compressedsensing and the last one is the core to reconstruct the original signal accurately. Researchers haveproposed many effective reconstruction algorithms, but these algorithms have some defects: such asrequesting that sparsity of signal is known, or imbalance between speed of convergence andstability of the algorithm, reconstruction speed is slower. After deeply and detailed analysis thesignal sparse and measurement-matrix, the paper attempts to suggest improvements in some widerapplied algorithms based on the introduction of adaptive factor to overcome these maindisadvantages of the existence algorithm and improve the quality of reconstruction signal. Theproposed adaptive reconstruction algorithms have higher reconstruction quality, better noise robustand much stabler when simulate, analysis and compare with the improvement of the beforealgorithm. |
