A Review Paper Between Open Source and Commercial SDK and Performance Comparisons of Face Matchers

Abstract

Biometric identification systems must be able to distinguish between individuals in various conditions. This paper presents a review paper on face matchers between commercial and open source SDK and techniques available for face recognition. This paper also presents experiments done in performance comparisons between a commercial face matcher, VeriLook 5.6 and an open source matcher OpenBR 0.5 from a set of images collected from those who work in the offices in various situations including the different poses, and various illuminants. The experiment results taken from those officers in NECTEC have shown that a commercial face matcher performs better but this requires proper set up of face extraction parameters to ensure the optimal results. Further improvements would require other face detection algorithms to enhance the performance of both free and commercial face matchers according to their own particular shortcomings.