Contourlet域方向子带稀疏表示的图像压缩感知

摘要

为了提高压缩感知中图像的稀疏表示性能,提出了一种Contourlet域方向子带稀疏表示的图像压缩感知算法.将图像Contourlet分解后的多个高频子带根据方向正交特点进行重组,采用随机高斯矩阵对重组后的子带分别进行测量,实现压缩采样；利用正交匹配追踪法重建各子带系数,并进行Contourlet反变换重构原图像.实验结果表明,在相同采样率下,算法重构图像的主观视觉效果和峰值信噪比都优于小波压缩感知算法.%To improve image sparse representation in compressed sensing,this paper proposed a new compressed sensing algorithm based on sparse representation using Contourlet directional subbands.Firstly,it decomposed images by Contourlet transform,and restructured directional subbands according to orthogonality,and then it used Gaussian random matrices to measure subbands and used orthogonal matching pursuit (OMP) algorithm to recover the subbands coefficients.Finally,it applied inverse Contourlet transform to reconstruct images.The experiments show that the proposed algorithm has better improvement in image visual effect and peak signal-to-noise ratio (PSNR).