Automatic Landmark Extraction from Three-dimensional Head Scan Data

摘要

This paper explores approaches to automatically identify anthropometric landmarks based on their geometric appearance from a 3-D scan of the human head. We investigated the geometric features of most commonly used landmarks of the head/face and related them to invariant surface characteristics such as mean and Gaussian curvatures and other first or second order derivatives. Based on the analysis of these features, we developed a number of methods and techniques to locate a set of landmarks from 3-D scan data.