Wet Face Recognition in Uncontrolled Scenario

Abstract

Research in face recognition has been much advanced in addressing several factors including pose, illumination and expression variations, facial occlusions etc. However, the adverse weather conditions such as higher humidity, unexpected rain, snow fall and fog have a greater impact on performance of an autonomous automated access control system. Therefore, we propose to work on wet face recognition. In this work, the impact of wet-face on face based intelligent access controlled system is studied. An effective approach is proposed for the same. Extensive experiments demonstrate the effectiveness of the proposed method in eliminating wet from face. Specially, the performance accuracy of access control system is impressive and achieves 95.72% and 97.23% recognition accuracy with single and two authentication factors respectively.