Gabor Wavelet-Based Appearance Models

There has been considerable research in the last several years based on the principles of Active Appearance Models (AAMs). AAM is a robust methodology for general image (object) descriptions that incorporates shape and texture information. In this work, we extend the basic AAMs by developing a new method for texture description for the application of human facial modeling and synthesizing. The premise is to develop a better texture based model for the face that incorporates specific facial information such as wrinkling, micro-features (e.g., moles, scares, freckles, etc.), and aging features (e.g., sagging, hollowing of checks, weight gain, etc.) This paper proposes a new improvement in texture synthesis-Gabor Wavelet-based Appearance Model. The experimental results demonstrate the potential of this approach for face-based applications.