基于偏微分方程的图像修复算法运行缓慢,且无法恢复纹理细节,实用性较差。基于地统计学思想,提出一种简单有效的基于各向异性插值模型的图像修复方法。实验表明该方法具有计算复杂度低和能够恢复图像纹理细节的优点,对于图像小区域划痕具有很好的修复效果,具有较高的实用价值。
Partial differential equation based image inpainting algorithms have disadvantage of high time consuming and are not practical. Based on the thinking of spatial statistic, this paper presented an inpainting algorithm using an anisotropic inter- polation model. Experiment results show that the algorithm has the advantages of low computation cost and texture preserved ability. The algorithm is suitable for restoring missing information in small regions such as scribed gaps and is practicable in ap- plications.