基于贝叶斯压缩感知的信号处理研究.pdfVIP

值息技术与健鹏也 基于贝叶斯压缩感知的信号处理研究 Research on 5ignaJProcessiηIgBased OIJ Bayesian Compressed Sensing 董亚芬* DONG Ya-fen 压缩感知 (Compressed Sensing) 技术通过信号采样与压缩过程同时进行的方式，解决了传统的信号处 理方式中信号存储传输空间较大、处理效字较低的问题.信号重构是压缩感知技术中的关键问题之一， 贝叶斯压缩感知 ( Bayes i an Compressed Sens ing) 算法通过稀疏贝叶斯回归模型中的相关向量机来解 决压缩感知中的信号重构问题。仿真实验结果表明，贝叶斯压缩感知算法能够以较小的相对误差重构出 房、始信号，性能比较优越。 关键词 贝叶斯压缩感知:稀疏信号:信号重构 Abstract The problems of large storage and transmission space of signals and low efficiency of signal proc~ssing in traditional signal procωing methods are solved by Compressed Sensing (CS) technology through carrying out the sampling process and the compression process simultaneously. Signal reconstruction is one of the key problems in compressed 5εnsing technology. Bayesian Compressed Sensing (BCS) algorithrn is based on spar优bayesian regression model in the relevance vector machine to solve the problem of signal reconstruction in compressed sensing. The simulation experiment results show that the Bayesian Compressed Sensing algorithrn can reconstruct the original signal with lower relative error, and the performance ofthe technology is superior. Key words Bayesian Compressed Sensing; Sparse Signal; Signal Reconstruction doi: 10. 3969/j. issn. 1672-9528.2016. 12.007 0 引言 1 压缩感知基本理论 信号在日常生活中无处不在，科技的进步离不开各种信 在压缩感知技术中，采样过程和压缩过程是同时进行的， 号处理技术。例如，对语音信号处理可以进行说话人识别、 无需遵循奈奎斯特来样定律，在接收端只需要采用适当的恢 说话人情感识别，对光信号进行处理可以帮助分析光通信网 复算法就可以恢复出原始信号 [ IJ 。压缩感知理论由于其优势 络、光信号图像，电信号的应用则更加广泛，遍及医疗、生 在信息理论、图像处理以及核磁共振等医学领域受到了广泛 物等领域。传统的信号处理技术必须遵循奈奎斯特采样定律，