考虑到contourlet变换的多尺度多方向性以及对二维图像具有比小波变换更好的稀疏表示特性,提出了一种基于contourlet变换的图像复原算法。算法采用边界优化的方法,通过类期望最大化算法在contourlet域进行迭代计算,并最终获得惩罚似然函数的最优解。实验结果表明,与传统的基于小波变换的同类图像复原算法相比,基于contourlet变换的复原算法在保持了较低的运算代价的同时,更好地保护了图像的边缘和细节信息,峰值信噪比有0.6 dB-0.8 dB左右的提高。

In consideration of the multi-scaled and multi-directional properties of contourlet transform and its better “sparse” representation of 2-D images,an image restoration algorithm based on contourlet transform is proposed. The algorithm adopts a bound optimization approach and is computed in the contourlet domain iteratively by an expectation-maximization-type (EM-type) method,and then reaches to the optimal result of the penalized likelihood function finally. The experimental results show that,compared with the traditional algorithms based on the wavelet transform,the proposed algorithm based on contourlet transform can protect the detail and edge information of the restored image much better,with the peak singnal-to-noise ratio (PSNR) improvement about 0.6 dB-0.8 dB,and at the same time,the computational efficiency is also retained.