A Combination of Shape and Texture Classifiers for a Face Verification System

In this paper, we present a general framework for a combination of shape (shape Trace transform-STT) and texture (masked Trace transform-MTT) classifiers based on the features derived from the Trace transform. The MTT offers “texture” representation which is used to reduce the within-class variance, while STT provides “shape” characteristics which helps us maximize the between-class variance. In addition, weighted Trace transform (WTT) identifies the tracing lines of the MTT which produce similar values irrespective of intraclass variations. Shape and texture are integrated by a classifier combination algorithm. Our system is evaluated with experiments on the XM2VTS database using 2,360 face images.