A new algorithm for 3D object representation and its application for human face verification

There are many papers published on the use of 2D images for object representation. The fundamental issue with these approaches lies in the fact that a conventional image maps 3D real world objects into a 2D plane. This causes loss of 3D information - which means loss of 3D features as well for further analysis. There have been attempts using 3D and 2.5D scanners in order to access range data from objects for further analysis. However, these systems require special scanners and associated hardware, and many a times the complete information that is obtained from such systems is not required for further analysis. In this paper, we propose a simple system that projects a light beam over an object of interest and using a simple optical camera a video is recorded. This enables us to have 2D data as acquired from the camera along with the 3D data that is derived out of the deviations in the scan pattern because of the non-planarity of the object of interest. We have limited our research to using 3D features only for object representation and extended the same logic for face verification. The results that we have obtained show that in spite of using only 3D features, the system is able to verify faces with varied expressions correctly on 75% of the test data. The system is robust to slight tilt in head angle and normal human expressions like smile, frown, grin etc.