基于样本块和BSCB模型的壁画裂缝修复方法  

Restoring Method of the Tomb Fracture Based on Sample Patch and BSCB Model

作　　者：师晓波 蔺素珍[1] 

机构地区：[1]中北大学计算机与控制工程学院,太原030051

出　　处：《科学技术与工程》2017年第35期98-104,共7页Science Technology and Engineering

基　　金：山西省重点研发计划项目(201603D321128);山西省自然科学基金(2013011017-4)资助

摘　　要：针对采用传统方法虚拟修复北齐墓葬壁画复杂裂缝效果较差的问题,以水泉梁北齐墓室壁画为例,提出了基于改进的BSCB(bertalmio sapiro caselles bellester)模型与样本块修复算法相结合的虚拟修复方法。首先直方图增强后用最大类间方差法(Otsu)获取裂缝;其次,结合使用连通域标记算法与开、闭运算分离裂缝;再采用改进的BSCB模型对所有裂缝进行修补;然后用基于样本块图像修复算法消除扩散修复痕迹;最后利用颜色提取、替换恢复前景色彩,获得裂缝最终修复结果。实验结果表明,该方法同时修复壁画纹理区域的大裂缝与相连于壁画结构的小裂缝时,既能保证修复区域的清晰度,又能连接破损边缘,还能恢复部分区域的色彩;相较于其他修复方法,在边缘连续性上至少提升15.90%。A virtual repair method was proposd based on the improved BSCB( bertalmio sapiro caselles bellester) model combined with the sample patch inpainting algorithm which aim to solve the problem that the effect of virtual repairing complex crack effect of the northern cathedral fresco is poor by the traditional method. First,cracks are obtained by the between-cluster variance method( Otsu) after histogram enhancing the original image. Second,cracks are separated combining connected domain labeling algorithm with the opening and closing operation. Third,all cracks are repaired by the modified BSCB model. Then a diffusion repair was removd based on sample patch image inpainting algorithm. Finally crack repair results are gained by extracting color and replacing foreground color.Experimental results show that our method restoration the big cracks of the texture area and the small cracks connected in the murals structure,which can not only ensure the clarity of the repair area and connect broken edge,but also restore the color of part areas. Compare with another algorithm,edge continuity upgrades at least 15. 90%.

关 键 词：墓室壁画 最大类间方差法 BSCB模型 样本块 虚拟修复 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]