三维颅面点对应方法研究

发布时间：2018-04-23 15:44

本文选题：颅面复原 + 颅面点对应　；参考：《西北大学》2017年硕士论文

【摘要】：颅骨面貌复原技术在考古学、法医学、人类学、刑事侦查等领域有着广泛应用。相比传统的颅面复原方法,基于统计理论的复原方法更加客观与科学,其复原结果更加接近于真实面貌。而建立颅骨与颅骨之间、面皮与面皮之间的点对应关系是基于统计的颅面复原方法的前提与关键所在。针对三维颅面模型间的点对应关系问题,本文进行了深入研究,主要研究工作及进展包括:1.实现了三维颅面模型数据重建与点对应预处理。主要包括颅面模型三维重建、法兰克福(Frankfurt)坐标统一、特征点定义及标定。通过三维重建,重构了单层的颅骨和面皮的三维网格模型;通过一系列预处理,实现了不同模型坐标系的归一化,且所有样本模型的特征点数目相同,为后续点对应算法的研究工作提供了良好的数据基础。2.提出了一种基于特征点变形与多尺度约束的三维颅面点对应方法。该方法主要考虑到不同模型特征点间的相对位置差异。首先通过基于特征点的径向基函数变形方法实现模型间的非刚性配准,使得模型间近似重合;然后建立多尺度约束下的点对应关系。实验结果表明,该算法有效提高了点对应的准确率。3.提出了一种基于体素模型与多几何特征约束的三维颅面点对应方法。该方法充分考虑到顶点微分属性对颅面模型表面凹凸性和复杂性的强力表达。首先建立颅面模型的体素模型,缩小对应点的搜索范围;然后计算顶点的微分属性,并在多几何特征的共同约束下确定点对应关系。实验结果表明,此方法有效提高了点对应结果的准确性,同时降低了时间复杂度。4.本文设计并完成了三维颅面点对应系统。其主要目的是建立模板颅面模型与待对应颅面模型之间的点对应关系,并对点对应结果进行了可视化。
[Abstract]:Skull face restoration technology has been widely used in archaeology, forensic science, anthropology, criminal investigation and other fields. Compared with the traditional craniofacial restoration method, the method based on statistical theory is more objective and scientific, and the restoration results are closer to the real appearance. Establishing the corresponding relationship between skull and skull and between facial skin and facial skin is the premise and key of the statistical method of craniofacial restoration. Aiming at the problem of the point correspondence between the three dimensional craniofacial models, this paper makes an in-depth study, the main research work and progress include: 1. Three-dimensional craniofacial model data reconstruction and point correspondence preprocessing. It mainly includes 3D reconstruction of craniofacial model, coordinate unification of Frankfurt (Frankfurt), definition and calibration of feature points. Through 3D reconstruction, the three-dimensional mesh model of single layer skull and facial skin is reconstructed, and the normalization of different model coordinates is realized through a series of preprocessing, and the number of feature points of all the sample models is the same. It provides a good data base. 2. 2. A 3D craniofacial point correspondence method based on feature point deformation and multi-scale constraints is proposed. This method mainly considers the relative position difference between different model feature points. Firstly, the non-rigid registration between the models is realized by the radial basis function deformation method based on the feature points, which makes the models approximate coincidence, and then the point correspondence relationship under the multi-scale constraints is established. Experimental results show that the algorithm can effectively improve the veracity of the corresponding points. A 3D craniofacial point correspondence method based on voxel model and multiple geometric feature constraints is proposed. This method takes full account of the strong expression of vertex differential attributes on the surface convexity and complexity of craniofacial model. Firstly, the voxel model of craniofacial model is established to reduce the search range of corresponding points, and then the differential attributes of vertices are calculated, and the corresponding relationship of points is determined under the constraint of multiple geometric features. The experimental results show that this method improves the veracity of the point correspondence and reduces the time complexity. 4. A three-dimensional craniofacial point correspondence system is designed and completed in this paper. The main purpose is to establish the point correspondence between the template craniofacial model and the corresponding craniofacial model, and to visualize the point correspondence results.
【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【相似文献】