Eye region based multibiometric fusion to mitigate the effects of body weight variations in face recognition

Face recognition has certain impediments due to alignment, illumination, facial expressions. Several techniques have been proposed to rectify these challenges. In recent years, many researchers have addressed challenges due to ageing, plastic surgery, twin identification, make-up and hairstyle. But, the impact of weight variation on face recognition has not been explored much. In contrary to other facial regions such as the cheek or chin area, the region near the human eye is not much affected due to the body weight changes. In this paper, we explore the use of eye region information to mitigate the effects and stabilize the performance of the biometric recognition system. To this extent, we propose a multi-algorithmic and multimodal fusion strategies to combine the information from eye region (left and right). The experiments carried out on the publicly available eWIT database indicates the improved recognition performance by 6.42% when benchmarked with commercial face recognition system.