Face Alignment Models

Abstract

In order to interpret images of faces (e.g., for recognition), it is important to have a model of the different ways that a face may appear. Though faces vary widely, changes can be broken down into two categories—changes in shape and changes in the texture (patterns of pixel values) across the face—that are largely due to differences between individuals, but also due to changes in expression, viewpoint and lighting conditions. In this chapter, we describe a powerful method of generating compact models of shape and texture variation, and describe two methods—the Active Shape Model (ASM) and Active Appearance Model (AAM)—that fit an appearance model to an unseen image of the face so that we can interpret its underlying properties (e.g., identity).