基于模糊阈值的回溯分段弱正交匹配追踪算法

为提高分段弱正交匹配追踪(Stagewise Weak Orthogonal Matching Pursuit,SWOMP)算法的重构质量,改进SWOMP算法的重构性能,提出一种基于模糊阈值的回溯分段弱正交匹配追踪(Backtracking Stagewise Weak Orthogonal Matching Pursuit Al-gorithm based on Fuzzy Threshold,FTB-SWOMP)算法.该算法首先通过采用SWOMP算法初始化选取一些大于门限的原子,在每次迭代过程中引入回溯,采用基于模糊阈值的回溯方式删掉个别错误原子,实现自适应的选取原子来更新支撑集,每次迭代的过程不断更新扩大支撑集,逐步逼近信号的稀疏度.实验结果表明,相同实验条件下,新算法能够以高概率恢复原始一维信号,且重构误差小;对于在二维图像上的应用,新算法仍保证重构时间短的优势,并比原算法具有更好的重构精度.

Backtracking stagewise weak orthogonal matching pursuit algorithm based on fuzzy threshold

For the purpose of improving the reconstruction effect and performance of stagewise weak orthogonal matching pursuit (SWOMP)algorithm,a backtracking stagewise weak orthogonal matching pursuit algorithm based on fuzzy threshold(FTB-SWOMP) was proposed.At beginning,SWOMP algorithm is adopted to select initial atoms selection in this new algorithm,then backtracking method based on fuzzy threshold is used to delete some wrong atoms in each iteration.The support set is adaptable updated while it is being enlarged gradually in every iteration,so it can approximate the sparsity of signal.The experimental results show that,under the same condition,the new algorithm recovers the original one dimension signal with high probability and small reconstruction error.For two-dimensional image signal,compared with SWOMP algorithm,the new algorithm has higher reconstruction effect and higher prac-ticability,which still has the advantage of high reconstruction speed.