CGV: Small: Digital Forensic Facial Reconstruction from Incomplete Datasets

在法医鉴定中，骨骼证据分析和面部重建在死者身份识别中起着重要的作用。在目前的实践中，有效的头骨处理和面部重建都是人工完成的。该项目研究3D几何分析和建模算法，以自动化和增强碎裂/不完整的颅骨修复和颅面重建，并倡导从手动处理向数字平台的潜在演变，以获得更高的效率、健壮性和客观性。在这些取证任务中需要解决的两个关键挑战性问题是几何恢复(从小碎片)和几何形状合成(具有复杂的几何和语义约束)。本项目研究有效的三维形状匹配和变换技术来解决这些问题。几何恢复通过可靠的局部匹配和对称引导来解决；几何合成通过非均匀的体积变形在非均匀的内层上施加给定的几何约束来探索。新的几何算法可用于许多计算机图形学和视觉任务中的不完全数据建模和分析。在这个项目中，计算机科学家正在与法医专家合作建立一个数字计算平台，并评估这些新的几何算法在法医头骨处理和颅面重建中的应用。该项目促进了法医执法、考古学、生物人类学和颅面矫形术中不完整的数据分析和重建。研究与教育相结合，将研究进展纳入现有和未来的课程，让RET/REU/K-12学生参与几何建模研究和图形/可视化系统开发；并吸引K-12和代表性不足的学生参加STEM教育。

In forensic investigations, the skeletal evidence analysis and facial reconstruction play important roles in identification of the decedent. In current practice, effective skull processing and facial reconstruction are accomplished manually. This project studies 3D geometric analysis and modeling algorithms to automate and augment fragmented/incomplete skull restoration and craniofacial reconstruction and advocates a potential evolution of manual processing to a digital platform for better efficiency, robustness, and objectivity. Two key challenging problems to solve in these forensic tasks are geometric restoration (from small fragmented pieces) and geometric shape synthesis (with complex geometric and semantic constraints). This project studies effective 3D shape matching and transformation techniques to tackle these problems. Geometric restoration is solved through reliable partial matching and symmetry guidance; geometric synthesis is explored via heterogeneous volumetric deformation that enforces given geometric constraints on non-uniform interior layers. The new geometric algorithms can benefit incomplete data modeling and analysis in many computer graphics and vision tasks. In this project, computer scientists are collaborating with forensic specialists to build a digital computational platform and evaluate the application of these new geometric algorithms in forensic skull processing and craniofacial reconstruction. This project facilitates incomplete data analysis and reconstruction in forensic law enforcement, archaeology, biological anthropology, and craniofacial orthopedics. The research and education are integrated by taking research advances into existing and future courses, involving RET/REU/K-12 students in geometric modeling research and graphics/visualization system development; and attracting K-12 and under-representative students into STEM education.