摘要
为提高压缩感知图像的重构质量,提出一种分段弱选择自适应正交匹配追踪算法(SWAMP)。结合维纳滤波器,采用光滑PL分块压缩感知恢复算法(BCS-SPL),提高重构效率。通过自适应设定阈值对残差与观察矩阵的相关性进行判定,使得不再受经验影响构建初始候选集。通过引入弱选择标准,使算法能自适应地更新支撑集,重构原信号。仿真结果表明,在相同的测试环境下,SWAMP算法在稀疏一维信号和图像二维信号方面整体优于其它同类算法,具有用时少稳定性高的特点。
A stage-wise weak selected adaptive orthogonal matching pursuit algorithm was proposed for improving the reconstruction quality of compressed sensing images.The BCS-SPL algorithm,which combined the Wiener filter with PL compressive sen-sing signal recover algorithm,improved the reconstruction efficiency.Adaptively setting threshold was introduced to build the initial candidate set by estimating the relevance between iterative residue and measurement matrix.A weak selection strategy was introduced to determine the number of atoms and candidate atoms by adaptively updating support set.Under the same condition,the simulation results show the superior reconstruction performance of SWAMP for1D sparse signal and2D images.It shows higher processing speeds and stability.
作者
王烈
罗文
秦伟萌
WANG Lie;LUO Wen;QIN Wei-meng(School of Computer and Electronics Information Engineering, Guangxi University, Nanning 530004, China)
出处
《计算机工程与设计》
北大核心
2018年第12期3767-3773,共7页
Computer Engineering and Design
基金
广西自然科学基金(2013GXNSFAA0019339)
关键词
压缩感知
信号重构
弱选择
自适应
正交匹配
compressive sensing (CS)
signal reconstruction
weak selection
adaptation
orthogonal matching
