关键词: 超分辨率, 最大后验概率, 正则化

Abstract:

Aiming at the restoration of blurred image, an improved adaptive super-resolution algorithm proposed which is based on morphological edge preserving operator. First， we define a morphology edge preserving operator,and the operator can keep adaptive adjustment as iteration.Secondly, applying the operator to the regularization during super-resolution reconstruction, so the constraint reconstruction is reinforced in the edge region and the regularization is emphasized in the smooth region. Experimental results demonstrate that the improved method preserves edges clearer and has better result in speed and robust for the regularization parameter.

Key words: super-resolution, maximum a posteriori probability（MAP）, regularized