电子学报
電子學報
전자학보
ACTA ELECTRONICA SINICA
2014年
6期
1061-1067
,共7页
田文飚%康健%张洋%芮国胜%张海波
田文飚%康健%張洋%芮國勝%張海波
전문표%강건%장양%예국성%장해파
压缩感知%去噪%自适应重构%卡尔曼滤波
壓縮感知%去譟%自適應重構%卡爾曼濾波
압축감지%거조%자괄응중구%잡이만려파
compressed sensing%denoising%adaptive reconstruction%Kalman filter
现有的贪婪迭代类压缩感知重构算法均基于最小二乘对信号进行波形估计,未考虑到可能将量测噪声引入信号估计的情况。针对以上不足,提出了一种基于线性Kalman滤波的压缩感知弱匹配去噪重构算法。该算法不需已知稀疏度先验,通过引入Kalman滤波,在最小均方误差准则下,每次迭代都获得最佳信号估计;并以弱匹配的方式同时筛选出有效的原子,并剔除冗余原子进而重构原信号。新算法继承了现有贪婪迭代类算法的有效性,同时避免了因噪声干扰或稀疏度未知导致的重构失败。理论分析和实验表明,新算法在同等条件下,重构性能优于现有典型贪婪迭代类算法,且其运算时间低于BPDN算法和同类的KFCS算法。
現有的貪婪迭代類壓縮感知重構算法均基于最小二乘對信號進行波形估計,未攷慮到可能將量測譟聲引入信號估計的情況。針對以上不足,提齣瞭一種基于線性Kalman濾波的壓縮感知弱匹配去譟重構算法。該算法不需已知稀疏度先驗,通過引入Kalman濾波,在最小均方誤差準則下,每次迭代都穫得最佳信號估計;併以弱匹配的方式同時篩選齣有效的原子,併剔除冗餘原子進而重構原信號。新算法繼承瞭現有貪婪迭代類算法的有效性,同時避免瞭因譟聲榦擾或稀疏度未知導緻的重構失敗。理論分析和實驗錶明,新算法在同等條件下,重構性能優于現有典型貪婪迭代類算法,且其運算時間低于BPDN算法和同類的KFCS算法。
현유적탐람질대류압축감지중구산법균기우최소이승대신호진행파형고계,미고필도가능장량측조성인입신호고계적정황。침대이상불족,제출료일충기우선성Kalman려파적압축감지약필배거조중구산법。해산법불수이지희소도선험,통과인입Kalman려파,재최소균방오차준칙하,매차질대도획득최가신호고계;병이약필배적방식동시사선출유효적원자,병척제용여원자진이중구원신호。신산법계승료현유탐람질대류산법적유효성,동시피면료인조성간우혹희소도미지도치적중구실패。이론분석화실험표명,신산법재동등조건하,중구성능우우현유전형탐람질대류산법,차기운산시간저우BPDN산법화동류적KFCS산법。
Almost all of the existing greedy iterative compressed sensing reconstruction algorithms estimate the signal by the method of least squares ,which introduces the measure noise into the signal estimation .Aiming at this problem ,a new weakly match-ing pursuit denoising recovery for compressed sensing based on Kalman filtering is proposed .The new algorithm does not need the sparse prior while it estimates the signal best for each iteration according to the minimum mean-square error criterion by Kalman fil-tering .Meanwhile ,weakly matching pursuit is used to sift the effective support set and pick out the redundancy to recover the origi-nal signal .The new algorithm is as effective as other greedy ones and is able to avoid recovery failure due to noise interference or unknown sparsity as well .The theoretical analysis and experimental simulation prove that the performance of the new algorithm is better than that of the existing greedy iterative reconstruction algorithms in the same condition .The operation time of the new algo-rithm is shorter than that of BPDN and the similar KFCS algorithm .