Abstract

3D Face recognition has been an area of interest among researchers for the past few decades especially in pattern recognition. The main advantage of 3D Face recognition is the availability of geometrical information of the face structure which is more or less unique for a subject. This paper focuses on the problems of person identification using 3D Face data. Use of unregistered 3D Face data for feature extraction significantly increases the operational speed of the system with huge database enrollment. In this work, unregistered Face data, i.e. both texture and depth is fed to a classifier in spectral representations of the same data. 2D Discrete Cosine Transform (DCT) is used here for the spectral representation. The face recognition accuracy obtained when the feature extractors are used individually is evaluated. The use of depth information alone in different spectral representation was not sufficient to increase the recognition rate. So a fusion of texture and depth statistical information of face is proposed. Application of statistical method seems to degrade the performance of the system when applied to texture data and was effective in the case of depth data. Fusion of the matching scores proves that the recognition accuracy can be improved significantly by fusion of scores of multiple representations. FRAV3D database is used for testing the algorithm.