面向压缩感知的稀疏度自适应图像重构算法研究

Sparsity Adaptive Image Reconstruction Algorithm for Compressed Sensing

自压缩感知理论(Compressed Sensing,CS)提出以来,重构算法的研究在CS技术中占据着重要地位,并受到了学者高度重视.针对目前重构算法在信号压缩采样中稀疏度未知这一缺点,提出一种稀疏度自适应的压缩采样匹配追踪算法(Sparsity Adaptive Compressive Sampling M atching Pursuit,SACo Sa M P).同时结合峰值信噪比(Peak Signal-to-Noise Ratio,PSNR)、重构误差概率(Reconstruction Error Possibility,REP)等指标衡量算法的图像重构性能,仿真结果表明:在测量矩阵满足有限等距性质(Restricted Isometry Property,RIP)的条件下,本文提出的算法具有自适应能力强,准确度高,图像重构效果佳等优点.

Since compressed sensing theory is proposed, the algorithm of image reconstruction plays an irreplaceable role in CS and a- rouses researchers＇ wide concern. A Sparsity Adaptive Compressive Sampling Matching Pursuit algorithm is proposed in order to tackle unknown sparsity of current greedy algorithms in compression sampling. And meanwhile, the performance of image reconstruction algorithm can be evaluated by making use of Peak Signal-to-Noise Ratio and Reconstruction Error Possibility. The simulation results indicate that the proposed algorithm has the following advantages of strong adaptability, high accuracy and amazing image reconstruction effects under meeting the condition of Restricted Isometry Property.