基于亚高斯随机投影的图像重建方法.pdfVIP

计算机研究与发展 ISSN 10002 1239PCN 112 1777PTP Journal of Computer Research and Development 5( 8) : 1 022 1 07, 2008 1, 2 1 1 方 红 章权兵 韦 穗 1 ( 安徽大学计算智能与信 处理教育部重点实验室 合肥 230039) 2 ( 合肥工业大学理学院 合肥 230009) ( luckymars@gmail1 com) A Method of Image Reconstruction Based on Sub2Gaussian Random Proj ection 1, 2 1 1 Fang H ong , Zhang Quanbing , and Wei Sui 1 ( Ministry of Education Key Laboratory of Intelligent Computing and Signal Processing , Anhui Uni ersity, Hef ei 230039) 2 ( Colleg eof Science, H ef ei Uni er sity of Technology, H ef ei 230009) Abstract In this paper, sub2Gaussian random projection is introduced into compressed sensing (CS) theory and two new kinds of CS measurement matr ix: sparse projection matrix and very sparse projection matrix are presented1 By the tail bounds for sub2Gaussian random projections, the proof of how these new matrices satisfy the necessary condition for CS measurement matrix is provided1 Then, it is expatiated that owing to their sparseness, new kinds of matrices greatly simplify the pr ojection operation dur ing image reconstruction, which simultaneously greatly improves the speed of reconstruction1 Further, it can be easily proved that Gaussian matrix and Bernoulli matrix ar e special matrices obeying sub2Gaussian random distribution, which indicates that new measur ement matrices extend the current results on CS measurement matrix1 Both the r esults of simulated and real experiments show that with a certain number of measurements, new matrices have good measurement effect and can acquir e exact reconstruction1 Finally, the comparison and analysis of reconstruction results respectively adopting