中文摘要：

为获得适用于不同模糊图像且简捷的图像盲复原方法，提出了一种稳健的从单幅模糊图像中求取模糊核并对图像去模糊的图像盲复原方法。根据模糊图像与非模糊图之间的边缘关系求模糊核，并在多尺度框架下针对各个子算法设定自适应参数，从而构建一个稳健的图像盲复原方法。对复原结果用4种无参考的图像质量评价方法的评价结果显示，本文方法在噪声和波纹抑制的基础上有更强的去模糊能力，参数调节简单，算法稳健，而且运算速度提高了3倍。

英文摘要：

In order to obtain a robust blind image restoration method that is applicable to different blurred images and without complicated parameter tuning, a robust blind image restoration method of extracting blur kernel from a single image and deblurring is proposed. In this paper, an approach of blur kernel esti- mation is presented according to the relationship of image edges between blurred images and nowblurred ones,and a robust blind image restoration method is constructed under the framework of nmlti-scale space with some adaptive parameters introduced in each sub-algorithm. A variety of no-reference image quality evaluation methods on the result show that the presented method has a better ＆blurring ability on the basis of suppression of noise and ringing with simpler parameter tuning, which is stable and steady,and its computation speed is enhanced by 3 times.