3D face recognition based on kinect depth data

Abstract

In this paper, a contour map human facial recognition algorithm is proposed to implement the three-dimensional (3D) face recognition with the Kinect Xbox One. Since the scale of 3D depth data collected from Kinect is tremendous, the face recognition process cannot be handled in real time. To improve the speed and accuracy of the recognition process, the proposed algorithm turns the 3D depth data to the two-dimensional (2D) contour map. Furthermore, due to the 3D depth data obtained by Kinect, there is no need of expensive, ponderous and slow 3D scanners. Ten male and female subjects were involved in the validation experiment and the results verify that the proposed algorithm was feasible for face recognition. In addition, compared with other methods, Eigenface, Local Binary Patterns (LBP) and Linear Discriminant Analysis (LDA), the proposed algorithm has the better security and reliability.