近几年来,贝叶斯压缩感知(BCS)技术得到了快速的发展并逐渐成为压缩感知领域的一项主流技术.该技术主要针对压缩感知中的重构部分,与传统的重构算法不同,其应用的是贝叶斯概率模型,而不是传统的1范数最小化模型.BCS的核心是相关向量机(RVM),但是,应用传统的RVM进行信号重构往往精度非常差.为了提高精度,文中提出了一种新的BCS技术:粒子群贝叶斯压缩感知(PSBCS).实验表明这种新的BCS技术在重构精度上大大超越了传统的BCS技术.%A new reconstruction metric for compressive sensing technique called the Bayesian compressive sensing ( BCS) was proposed in the recent years. It considers the reconstruction process as the Bayesian model rather than the traditional l1 norm sparsity model. In BCS, the so called the relevance vector machine ( RVM) is used, which have better performance than the l1, norm sparsity model. However, the accuracy of the conventional BCS is very low, which means its performance is highly dependent on the signal. To enhance accuracy of the conventional BCS, a new kind of modified BCS named the particle swarm Bayesian compressive sensing (PSBCS) is proposed in this paper. The experiments show that the PSBCS outperforms the conventional BCS and other reconstruction metrics for its high accuracy on signal reconstruction.
展开▼
