Abstract: Among the non-blind restoration algorithms for image processing, the accuracy of the point spread function will directly affect the quality of the image restoration. There are two traditional methods to estimate the PSF, one is direct detection, the other is indirect estimation. While two methods need large amount of dates and have hard calculation. Aiming at these problems, this paper uses differential evolution algorithm to estimate the point spread function of image degradation from the view of global optimization. Experimental results illustrate that differential evolution algorithm has good performance, and laying a good foundation for designing the restoration filter next step.

Key words: point spread function, differential evolution algorithm, non-blind restoration algorithms

摘要： 在图像处理的非盲复原算法中，对点扩散函数PSF（Point Spread Function）的估计是否准确，将直接影响图像复原质量的好坏。传统的估计方法有直接测量法和间接估计法，但往往数据量大，计算繁琐，针对这一问题，从全局寻优的角度出发，提出利用差分进化算法来估计点扩散函数。实验结果表明差分进化方法可以取得很好的效果，为下一步设计恢复滤波器奠定了基础。

关键词: 点扩展函数, 差分进化算法, 非盲复原算法