基于最小均方误差估计和稀疏性先验的图像去噪

An image denoising algorithm using MMSE estimation with a sparsity prior

提出一种基于稀疏表达理论和最小均方误差估计的图像去噪算法,主要内容包括:在贝叶斯复原框架下,根据图像在冗余字典下的稀疏表达模型,建立原始图像表达系数的最小均方误差复原方程;利用随机正交匹配追踪算法,研究复原方程的数值求解算法,对图像表达系数进行近似求解,进而对原始图像进行恢复.对一组标准测试图像的仿真实验表明,提出的算法能够较好地去除图像中的噪声,并且复原图像具有较好的主观视觉质量和较高的峰值信噪比客观评价指标.

The paper proposed an image denoising algorithm based on sparse representation and minimum mean square error estimation (MMSE).The main research contents included: under the Bayesian restoration framework,we exploited the sparse representation model of image to established the MMSE equation of the representation vector of the original image, and researched the numerical algorithm to approximately solve of the MMSE estimation equation by the random orthonormal matching pursuit (OMP)algorithm.The experimental result according to a group of standard test image showed that the proposed method could effectively removed the addictive noise,and performed good subject visual quality and peak signal to noise ratio (PSNR)values of the restored image.