Ascertaining Human Identity In Night Environments

Understanding patterns of human activity from the fusion of multimodal sensor surveillance sources is an important capability. Most related research emphasizes improvement in the performance of biometric systems in controlled conditions characterized by suitable lighting and favorable acquisition distances. However, the need for monitoring humans in night environments is of equal if not greater importance. This chapter will present techniques for the extraction, processing and matching of biometrics under adverse night conditions in the presence of either natural or artificial illumination. Our work includes capture, analysis and evaluation of a broad range of electromagnetic bands suitable for night-time image acquisition, including visible light, near infrared (IR), extended near IR and thermal IR. We develop algorithms for human detection and tracking from night-time imagery at ranges between 5 and 200 meters. Identification algorithms include face, iris, and gait recognition, supplemented by soft biometric features. Our preliminary research indicates the challenges in performing human identification in night-time environments.