文献类型：期刊文献

英文题名：Image reconstruction based on improved block compressed sensing

作者：Du, Hong[1];Lin, Huixian[1]

机构：[1]Guangdong Ocean Univ, Fac Math & Comp Sci, Zhanjiang 524088, Peoples R China

年份：2022

卷号：41

期号：1

外文期刊名：COMPUTATIONAL & APPLIED MATHEMATICS

收录：SCI-EXPANDED(收录号：WOS:000728575700001)、、EI(收录号：20224112884995)、Scopus(收录号：2-s2.0-85120938130)、WOS

语种：英文

外文关键词：Block compressed sensing; Sparse matrix; Mallat algorithm; Db2 wavelet; Image reconstruction

外文摘要：Compressed sensing (CS) technique can sample and compress simultaneously. Since an image contains a huge amount of information, block CS (BCS) technique has appeared. This technique can divide image signals into non-overlapping sub-blocks and all the sub-blocks are processed separately. In classical BCS, translations of coefficients may cause signal losses when constructing sparse matrices. Therefore, reconstructed images are unsatisfactory at low sparsity. In this paper, we propose an improved BCS (IBCS). Our implementation is based on Mallat reconstruction algorithm to construct a non-square sparse matrix, it can retain more information of original signals. The reconstruction quality is more stable at different sparsity. Two experiments demonstrate that the reconstruction quality of the proposed IBCS is better than that of CBCS at lower sparsity and the implementation cost is reduced.