基于Contourlet变换的图像分块压缩传感重构算法

摘要

随着压缩传感的广泛应用,其在图像重构方面的优势得以体现.Contourlet凭借其在图像轮廓和纹理方面的出色表现,成为最受欢迎的方向变换之一.文中提出了一种采用Contourlet变换作为稀疏基,并且运用图像分块采样技术,结合平滑投影Landweber迭代的图像重构算法.实验表明,相较于采用DCT和DWT作为稀疏基的图像重构算法,提出的算法使得重构图像的质量有所提高,尤其是在低采样率的情况下.%With the development of compressed sensing theory,it shows the advantage in reconstruction of images.Contourlet transform is one of the most popular directional transforms because of its brilliant performance in preserving features such as contours and textures.This paper proposes an image reconstruction in which the Contourlet transform is sparsified and using block-based random image sampling is coupled with a smoothed projected Landweber iteration.The experimental results show that compared with the sparse representation of the DCT or DWT,the propsed algorithm improves the quality of reconstructed image,especially is suitable for the case of low sampling rate.