Face recognition using two different 3D sensors

Theoretical and technical advances in 3D data capture techniques present newer disciplines such as 3D face recognition for biometric identification. This paper presents a 3D face recognition system based on range images under pose varying environments. A range image from a structured light system has been adopted as input data and 3D laser scanned data form a database. 3D head pose estimation is performed to estimate the head pose of input data with extracted facial features. In addition, we generate range images from the 3D database according to estimated parameters in order to perform face recognition. We compare an input range image based on the correlation method and Principle Component Analysis (PCA) with generated range images. Experimental results demonstrate that the proposed method provides 93% recognition rate and can be adopted for 3D biometric identification systems.