基于内表面特征的碗状碎块匹配方法

Matching Method for Bowl-Shaped Broken Pieces Based on the Features of Inner Surfaces

传统的文物碎块拼接方法会因为断裂曲面丰富的几何信息产生过分割的问题,而碗状碎块不同于其他文物碎块,其特点在于有鲜明的内表面特征可以利用,因此提出一种基于内表面特征的碗状碎块匹配方法。该方法首先结合点云占比和平滑度提取碗状碎块的内表面,接着构造类抛物线模型描述边缘点的特征圆曲率,并建立扩展高斯球以表征碗状碎块的全局特征,然后定义相似度从碎块集中筛选出最佳匹配的一对碎块并进行碎块的粗匹配,最后结合ICP精匹配算法完成碎块集的重组。实验结果表明基于内表面特征的碗状碎块匹配方法可以有效地完成碎块集的整体匹配,从而避免过分割的问题,并且碎块筛选的准确率和效率均高于传统方法。

The traditional matching method of cultural relic fragments will have the problem of over segmentation because of the rich geometric information of fracture surfaces,while the bowl-shaped broken pieces are different from other cultural relic fragments,which is characterized by the distinct internal surface features that can be used.Therefore,a matching method for bowlshaped broken pieces based on the features of inner surfaces is proposed.Firstly,this method extracts the inner surface of bowlshaped broken pieces by combing point cloud proportion and smoothness.Then,a likeness parabolic model is constructed to obtain the characteristic circular curvature of boundary points,so that the extended Gauss sphere is established based on the features of inner surfaces to represent the global features of bowl-shaped broken pieces.Then,the similarity is defined to screen the best matching pair of broken pieces and perform the rough matching of broken pieces.Finally,the reassembly of broken pieces set is completed with combining the ICP fine matching algorithm.The experimental results show that the matching method for bowlshaped broken pieces based on the features of inner surfaces can effectively complete the whole matching and avoid the problem of over segmentation.The accuracy and efficiency of screening are also higher than the traditional methods.