Forensic Facial Reconstruction from Skeletal Remains

Abstract

The identity of a skull in forensic is of critical importance. Forensic facial reconstruction is the reproduction of the lost or unknown facial features of an individual. In this paper, we propose the automation of the reconstruction process. For a given skull, a data-driven 3D generative model of the face is constructed using a database of CT head scans. The reconstruction can be constrained based on prior knowledge of parameters such as bone thickness measurements, cranial landmark distance measurements and demographics (age, weight, height, and BMI). The CT scan slices are segmented and a 3D model skull of 2D slices is generated with the help of Marching Cubes Algorithm. The 66 Landmark points are then calculated using Active Shape Models and PCA algorithm and placed on the skull. These Landmark points act as references for tissue generation. The facial soft tissue thickness is measured and estimated at the 66 craniometric landmarks used in forensic facial reconstruction. The skin mesh is generated using Delaunay automatic triangulation method. The performance of this model is then measured using RSME technique. The aim of this study is to develop a combination of techniques and algorithms to give the most accurate and efficient results.