改进的区域生长算法在寺观壁画脱落病害标定中的应用

The Application of Improved Region Growing Algorithm for the Automatic Calibration of Shedding Disease on Temple Murals

针对开化寺宋代寺观壁画存在的脱落病害侵蚀问题,提出一种融合阈值分割的改进的区域生长算法,自动标定壁画的脱落病害.首先分析脱落区域的颜色特征,通过阈值分割标注疑似脱落点并以这些点为种子点进行区域生长,扩展脱落区域,计算颜色掩码;然后在YcbCr、HSV颜色空间分析脱落区域的亮度、色度、饱和度特征,通过阈值分割得到脱落区域的亮度、色度、饱和度掩码,并将各个特征掩码进行融合;最后将融合得到的脱落区域掩码与原图进行加运算,实现脱落病害的标定.与现有壁画病害标定算法进行对比,结果表明本文提出的标定算法的标定效果更好,为古代壁画的虚拟和实际修复奠定了良好的基础.

Aiming at the problems of shedding disease of Kaihua Temple murals in Song Dynasty, an improved region growing algorithm fusing threshold segmentation was proposed to calibrate shedding disease for murals automatically. First, the color features of shedding areas were analyzed, the suspected shedding points which were taken regard as seed points to make region growth were marked by threshold segmentation,and the color mask was calculated. Second, brightness, chroma and saturation features of the shedding areas were analyzed in the color space of YcbCr or HSV and the masks of brightness, chroma and saturation were extracted using threshold segmentation. Then each feature mask is fused. Finally, the mask of shedding area after fusing was added to the original mural to achieve the calibration results of the shedding disease.The experiment results show that the automatic calibration algorithm proposed in this paper has a better calibration effect, which lays a solid foundation for the virtual and real restoration of the ancient murals.