V4 N2 Paper 1
Annals of the MS in Computer Science and Information Systems at UNC Wilmington
Fall 2010

Constructing a 3D morphable face from a single picture  

Seiken Higashionna

Committee

Eric Patterson (chair)
Ulku Clark
Karl Ricanek

Abstract

This paper presents the construction of 3D face models using a correspondence established between 2D Active Appearance Models (AAM) and 3D Morphable Models (3DMM).The hypothesis is that a 3D face model may be generated via a 2D-to-3D corresponds between Active Appearance parameters and 3D Morphable Model parameters. There are a few different approaches that are currently used to construct 3D morphable face models. This work, in contrast though, is geared toward simplicity and speed in establishing a rough likeness that could be applied in a variety of 3D face software such as games, animation, etc. In theory, this technique will construct a 3D morphable face significantly faster compared to one of the primary techniques thus far, Blanz and Vetter’s approach described in their previous work. The quality of generated 3D models is analyzed subjectively and quantitatively to conclude that this has merit.

download (pdf)

Recommended Citation: Higashionna, S., Patterson, E, Clark, U., Ricanek, K. (2010) Constructing a 3D morphable face from a single picture. Annals of the Master of Science in Computer Science and Information Systems at UNC Wilmington, 4(2) paper 1. http://csbapp.uncw.edu/data/mscsis/full.aspx.

V4 N2 Paper 1
Annals of the MS in Computer Science and Information Systems at UNC Wilmington
Fall 2010