基于改进优先级的自适应图像修复算法

Adaptive image inpainting algorithm with modified priority

文章基于Criminisi算法,提出了改进优先级的自适应样本块匹配算法,在Criminisi算法的优先级中增加了信息项,使图像的修复顺序更加合理;针对以往修复图像时样本块大小固定不变,造成图像的不合理结构累积,部分区域的局部结构丢失等不足,提出一种改进的自适应样本块匹配方法,其核心思想是通过局部结构特征,确定修复样本块大小。实验结果表明,所提算法对许多图像都能取得较好的修复效果。

In this paper ,an adaptive sample patch matching algorithm with a modified priority is given based on the Criminisi algorithm .An information term is added to the priority in Criminisi algorithm to make the image inpainting order more reasonable .In addition ,for avoiding the problem that the sample patch size of previous restoration is fixed ,which will cause the accumulation of irrational structure of the image and the loss of the local structure of some region ,an improved adaptive sample patch matching method is given to determine the sample patch size by local structure features .Experi-mental results show that the algorithm given in the paper has good inpainting results .