A new image reconstruction algorithm of block compressed sensing
Qingshan She,Ping Xu,Zhizeng Luo,Dongliang Liu
DOI: https://doi.org/10.3969/j.issn.1001-0505.2011.S1.007
2011-01-01
Abstract:Due to the fact that the traditional compressed sensing (CS) methods need to access the whole image at once in the random sampling process, which results in more storage and reconstruction time, a novel reconstruction algorithm based on block CS is proposed. First, the image is divided into a series of subblocks, and each subblock is concatenated vectors column-wise to represent the original signal. Secondly, the measurement values of the signal are taken by projecting onto a measurement matrix after sparse transformation, and then the recovered signal can be obtained from these measurement values using some optimization methods. Thirdly, through subblock activity classification, the subblock boundaries with different activity indices are processed with different post-filtering methods. Finally, magnetic resonance imaging (MRI) images are used to validate the proposed algorithm. Experimental results show that the proposed approach can decrease reconstruction time and reduce blocking artifacts and preserve textures and edges.