Computer-generated face models
Bernard Tiddeman · Cambridge University Press eBooks · 2012
Introduction The rapid progress in digital image capture and computer processing power has provided new opportunities for the analysis and synthesis of facial shape and appearance. New algorithms have been developed to exploit these improvements and have found applications in health, security and forensics. In this chapter some of the main techniques that have been developed to capture, model and process facial image data are reviewed. Data capture Photography The simplest technique for capturing facial image data is photography. There have been rapid improvements in the resolution of digital cameras, so that the quality of direct digital capture equals or even exceeds chemical film. While two-dimensional (2D) images present the simplest method for facial image capture, 2D data present a number of problems when it comes to analysing and processing the data. Faces are inherently 3D in nature and the projection to 2D can introduce variation in the appearance that is not intrinsic to the faces themselves. Even small changes in the facial orientation can produce a vastly different projected 2D facial shape. Changes in lighting are also difficult to model and remove using purely 2D methods.