Xiong-yong Zhu¹, Shun-dao Xie^2,3, Guo-ming Chen¹, Liang Xue¹, Wen-fang Wu², Hong-zhou Tan^2,3, ¹Guangdong University of Education Guangzhou, China, ²Sun Yat-Sen University Guangzhou, China and ³SYSU-CMU Shunde International Joint Research Institute, China

Traditional methods to signal acquisition need to collect large amounts of redundant data, and then compress the data to extract useful information, which is inefficient and requires large amount of storage resources. Compressed sensing (CS) can avoid sampling the redundant data; it obtains the discrete signals at the sampling rate that is lower than the Nyquist sampling rate, and reconstructs the original signal with high probability. Based on CS, Block Stagewise Regularized Orthogonal Matching Pursuit (StROMP) is proposed in this paper to reconstruct images. Simulation results show that the proposed algorithm can effectively reduce the required storage storages and computational complexity, which improves the quality of reconstructed images in the premise of ensuring a shorter reconstruction time.

Compressive Sensing; Matching Pursuit; Image Reconstruction