基于SWOMP分段回溯的压缩感知改进算法

Improved compression sensing algorithm based on SWOMP stagewise backtralcing

为解决压缩感知中阶段性弱正交匹配寻踪算法(SWOMP)重建精度低的问题,提出一种分段回溯弱选择正交匹配跟踪算法(SCWOMP)。该算法在SWOMP算法基础上加以改进,首先采用算术阈值策略和最大相关列两种状态,根据具体阶段选择不同标准提高所选索引的精度,然后引入一个回溯步骤灵活地删除以前处理中错误选择的一些索引,以较低的复杂度获得较好的重建性能。仿真结果表明,在相同条件下,相较于SWOMP算法,SCWOMP算法可以得到更好的重建效果和较准确的重建概率。

This paper proposes an effective algorithm of piecewise stage backtracking weak orthogonal matching tracking(SCWOMP)to address the lower reconstruction accuracy of stage weak orthogonal matching pursuit(SWOMP)algorithm in compressed sensing,and to overcome the inaccuracy of index selection.The algorithm,improved on the basis of SWOMP algorithm works by firstly improving the precision of the selected index by applying the arithmetic threshold strategy and maximum correlation column and selecting different criteria according to specific stages and then achieving a better reconstruction performance with a lower complexity by introducing a backtracking step to flexibly delete some wrong indexes in the previous processing.The simulation results show that with the same conditions,SCWOMP algorithm boasts a better reconstruction effect and higher accurate reconstruction probability than SWOMP algorithm.