基于广义高斯分布的最大后验概率图像复原算法

Algorithm of Image Restoration Maximum Posteriori Probability based on Generalized Gaussian distribution

针对传统图像复原算法在减轻图像噪声和去模糊方面存在较大振铃现象的问题,提出了一种基于广义高斯分布模型的最大后验概率图像复原算法。该算法将最大后验概率模型和广义高斯分布相结合,利用了广义高斯分布可以模仿多数噪声模型的优势。与传统的最大似然算法相比较,所提出的算法不仅能有效地改善噪声,而且还能减轻由复原过程造成的纹波现象。

Aimed at the puzzle of larger ringing phenomena existed in minimizing image noise and deblurring for the algorithm of traditional image restoration, the paper proposed a sort of algorithm of image restoration for maximum posteriori probability based on generalized Gaussian distribution. The algorithm combined the maximum posteriori probability model with generalized Gaussian distribution, it could simulate the advantage of the most noise model by means of generalized Gaussian distribution. Compared with the conventional maximum-likelihood method, the proposed algorithm can not only suppress the noise effectively, but also alleviate the ripple wave phenomenon caused by restoration process.