中文摘要：

通过改进Favaro等人提出的迭代求解热扩散方程的方法,研究了由散焦图像恢复物体深度的问题.首先将两幅散焦图像按照不同的相对模糊程度划分为两个区域,证明了这两个区域的边界和图像的边缘有很高的一致性;然后通过强化图像的边缘,使这两个区域的边界在迭代求解热扩散方程的过程中更容易确定;再根据确定的U＋和U-的边界较好地求解了物体深度;最后通过模拟实验及实际图像实验对该方法进行了验证.模拟实验结果表明,文中方法具有较好的鲁棒性;实际算例分析表明,文中方法的恢复效果优于Favaro的方法.

英文摘要：

This paper deals with the depth recovery from defocused images by improving Favaro＇s iteration method based on thermal diffusion equations.In the investigation,two defocused images are respectively divided into two regions according to the relative blur,whose borders are highly consistent with the image edge and can be easily determined by strengthening the image edge during the solving of the thermal diffusion equation via iteration.The depth can be accurately obtained according to the determined U＋ and U-boundaries.The effectiveness of the proposed method is then validated by simulation and real image experiment.Simulated results show that the proposed method is of high robustness,and the real-image experimental results show that the method is superior to Favaro＇s method in terms of depth recovery.