唐墓室壁画裂缝的自动虚拟修复方法

裂缝是壁画中存在的一种常见的病害，为了去除壁画中的裂缝提出一种对壁画中裂缝进行自动虚拟修复的方法。对图像进行高帽变换提取裂缝信息，对变换后的图像进行阈值分割得到裂缝的二值图像。根据裂缝的颜色和结构信息利用基于连通域度量的方法去除虚假目标，实现自动识别和标注。利用基于样本的数字图像修复技术对壁画中裂缝进行自动修复，在修复中对样本区域进行重构大大缩短了修复的时间，提高了修复效率。实验仿真表明该方法能够有效去除壁画中影响人眼视觉的较大裂缝，能够实现对壁画中裂缝的自动虚拟修复。

Virtual restoration method of cracks in mural

In order to remove the cracks in the mural, a virtual restoration of the cracks has been introduced. The cracks information is extracted by using top-hat transformation, and then its binary image is got by threshold segmentation. According to the color and structural information of cracks, the false cracks are removed by measuring the connected domain, so as to realize the automatic identification and label. The cracks are restored with example-based inpainting approach. By reconstructing the sample area, the repair time of the damage region has been greatly shortened, and the repair efficiency has been improved. The simulation experiments show that the large cracks can be removed and restored automatically.