Shape index based 3D face recognition using photometric stereo

There are various 3D reconstruction methods from 2D images. One of them is photometric stereo. This paper proposes simple and efficient 3D face recognition system using photometric stereo. Test 3D data and reference 3D data have been reconstructed using photometric stereo. Proposed face recognition algorithm has been tested with 38 people. 10 image groups are organized with combination of three different 2D images with different illumination directions and 3D test face data has been produced using these three 2D images. Reference 3D data has been created using 64 2D images that have been taken with different illumination directions. In recognition process, shape index maps have been calculated and then minimum differences between shape index maps have been analyzed. Results of our approach and other methods have been compared.