The damaged regions of Thangka images need to be segmented before repairing by digital technology. A segmentation algorithm is proposed taking the features of the faded regions in consideration. Stepl, binary processing and block processing are done in the gray image of damaged image, then AND operation on the two images after their processing are made, which leads to result 1. Step2, the gray image, covered with white regions in result 1 by the maximum variance method and the segmentation method based on mathematical morphology, will be further segmented, and then result 2 and result 3 are obtained. Step3, result 4 is gained, as a result of the combination of result 2 and result 3. Step4, the best effective image from result 1—4 will be selected, after removing the small-scaled regions and the false damaged regions, the final segmentation result will be achieved. Experimental results show that this algorithm has effective segmentation about the faded regions of Thangka and certain robustness.%采用数字化技术修复唐卡需先对唐卡破损区域进行分割,针对唐卡褪色区域的特点,提出一种分割算法.首先对破损图像的灰度图进行二值化和分块处理,并对处理后的2种图像进行与操作得到分割结果1；再对分割结果1中白色区域覆盖的灰度图利用最大类间分割法和基于数学形态学的分割法进行分割,得到分割结果2和分割结果3；合并分割结果2,3得到分割结果4；从分割结果1～4中选择分割效果最佳的图像,去除小面积破损区域和伪破损区域得到最终的分割结果.实验结果表明,该算法对唐卡褪色区域的分割效果较好,具有一定的鲁棒性.
展开▼
