摘要： 基于多种稀疏变换基和观测矩阵的组合，采用正交匹配追踪算法对图像进行重建。在分析各种组合重建效果的基础上，提出一种图像快速重建方法，对图像进行一级小波分解，提取出近似分量子图像，运用压缩感知技术对其进行恢复，综合细节分量和恢复出的近似分量进行小波逆变换，得到重建图像。实验结果表明，该方法在相同的观测值条件下，能减少算法运行时间，提高重建图像质量。

关键词: 压缩感知, 稀疏表示, 小波分解, 图像重建, 正交匹配追踪

Abstract: Orthogonal Matching Pursuit(OMP) algorithm is used to reconstruct images based on the combinations of several common sparse transform bases and measurement matrices. This paper analyzes and compares the reconstruction results with above various combinations. And on this basis, a fast image reconstruction method is proposed. A multi-scale wavelet decomposition is used to extract the approximate coefficients from the image. It uses compressed sensing method to recover these approximate coefficients, and the reconstructed images are obtained with inverse wavelet transform based on detail coefficients and recovered approximate coefficients. Experimental results show that with the same number of measurement values, the run time of this method and the quality of the reconstructive images have great improvements.

Key words: compressive sensing, sparse representation, wavelet decomposition, image reconstruction, Orthogonal Matching Pursuit(OMP)

中图分类号: 

• TP393