3D facial model analysis for clinical medicine

Abstract

This thesis aims to investigate facial paralysis diagnosis and facial highlight features based on 2D and 3D facial models. First, an automated objective asymmetry grading system is developed for facial paralysis diagnosis. The system combines observations and clinical assessments of the patients for motion dysfunction. Higher order surface properties for 3D model analysis are used. Facial symmetry grading is carried out based on registration result of the original and mirror facial meshes by the iterated closest-point algorithm. The noise injected artificial neural networks (ANNs) in classification were implemented. Compared with standard ANNs, the performance of the system are significantly improved. Second, the highlight feature patterns of natural faces are explored as a planning aid for plastic surgery. Different from reported studies on attractive face patterns, which have mainly based their criteria on facial profile, this study determines the position and shape of the highlights of faces across race and gender.