Revising regularisation with linear approximation term for compressive sensing improvement
Revising regularisation with linear approximation term for compressive sensing improvement
复制标题
使用线性近似项修改正则化以改进压缩感知
DOI：
10.1049/el.2018.8019
复制
发表时间：
2019
影响因子：
1.1
通讯作者：
Wang Shidong
中科院分区：
文献类型：
--
作者：
Chen Zan;Hou Xingsong;Shao Ling;Wang Shidong
在这封信中，作者提出了一种新的修正正则化，以提高压缩感知（CS）重建的性能。他们假设一个特定的正则化项不足以容纳CS的先验信息，而它可以通过进一步施加线性近似项来改善。他们还证明，修订后的正则化是实质上等同于CS预处理方法。他们对各种CS算法进行了广泛的实验，这些实验表明了他们修改后的正则化的有效性。
In this Letter, the authors propose a novel revised regularisation to improve the performance of compressive sensing (CS) reconstruction. They suppose that a specific regularisation term is insufficient to accommodate the prior information of CS while it can be improved by further imposing a linear approximation term. They also prove that the revised regularisation is substantially equivalent to the CS preprocessing methods. They conduct extensive experiments on various CS algorithms, which show the effectiveness of their revised regularisation.