分块稀疏信号1-bit压缩感知重建方法

发布时间：2018-05-21 06:51

本文选题：-bit压缩感知 + 变分贝叶斯推断　；参考：《物理学报》2017年18期

【摘要】：1-bit压缩感知理论指出:对稀疏信号进行少量线性投影并对投影信号进行1-bit量化,该1-bit信号包含足够的信息,从而能对原始信号进行高精度重建.然而,当信号难以进行稀疏表达时,传统1-bit压缩感知算法无法精确重建原始信号.前期研究表明,分块稀疏模型作为一种特殊的结构型稀疏模型,对于难以用传统稀疏模型进行表达的信号具有较好的表达作用.本文提出了一种针对分块稀疏信号的1-bit压缩感知重建方法,该方法利用分块稀疏的统计特性对信号进行数学建模,通过变分贝叶斯推断方法进行信号重建并在光电容积脉搏波(photoplethysmography)信号上进行了实验验证.实验结果表明,与现有1-bit压缩感知重建方法相比,本文方法重建精度更高,且收敛速度更快.
[Abstract]:The theory of 1-bit compression sensing points out that a small amount of linear projection of sparse signal and 1-bit quantization of the projected signal are carried out. The 1-bit signal contains enough information to reconstruct the original signal with high accuracy. However, when the signal is difficult to be expressed sparsely, the traditional 1-bit compression sensing algorithm can not accurately reconstruct the original signal. Previous studies have shown that as a special structural sparse model, block sparse model can well express signals that are difficult to express by traditional sparse model. In this paper, a 1-bit compression perceptual reconstruction method for block sparse signals is proposed, which makes use of the statistical characteristics of block sparse to model the signals. The signal is reconstructed by variational Bayesian inference and verified on photoplethysmography signal by photovolumic pulse wave. The experimental results show that the proposed method has higher accuracy and faster convergence speed than the existing 1-bit compression perceptual reconstruction methods.
【作者单位】：天津大学电气自动化与信息工程学院;立命馆大学机器人系;
【基金】：国家自然科学基金(批准号:61401303,51578189)资助的课题~~
【分类号】：TN911.7

【相似文献】