基于压缩感知的图像重构算法研究  

Image Reconstruction Algorithm Based on Block Sparse Matrix

作　　者：朱明明 黄星 Zhu Mingming;Huang Xing(Department of Physics and Engineering Technology,Guilin Normal College,Guilin,Guangxi 541199,China)

机构地区：[1]桂林师范高等专科学校物理与工程技术系,广西桂林541199

出　　处：《桂林师范高等专科学校学报》2021年第1期92-96,共5页Journal of Guilin Normal College

摘　　要：压缩感知重构图像时,重构图像的失真率会随着稀疏度的增大而增大。文章基于压缩感知的正交匹配追踪(OMP)重构算法,引入离散余弦变换(DCT)图像分块,降低由稀疏度增大对图像重构的影响。仿真结果表明:在相同的稀疏度条件下,改进重构算法重构图像的失真率要低于传统的OMP算法。在最佳分块数的条件下,改进重构算法PSNR值要比OMP高出1.5-3.2dB左右。When compressed sensing(CS)reconstructs the image,the distortion rate of the reconstructed image will also increase with the rise of sparsity in the reconstruction process.Based on the Orthogonal Matching Pursuit(OMP)reconstruction algorithm,Discrete Cosine Transform(DCT)image block is introduced to reduce the effect of sparseness increasing on image reconstruction.The simulation results indicate that under the same sparsity condition,the reconstruction algorithm has a lower distortion than the traditional OMP algorithm and under the condition of the optimal block numbers,its PSNR value is about 1.5-3.2dB higher than OMP.

关 键 词：压缩感知 稀疏度 失真率 正交匹配追踪 离散余弦变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]