Image processing algorithms underpinning iris and facial recognition systems
Christine E. Connolly · Sensor Review · 2006
Purpose To study the mathematical image coding approaches used in two types of biometric systems, and the physical nature of those biometrics. Design/methodology/approach Gives details of algorithms used to encode data from images in established and new automatic iris recognition systems. Then examines face recognition techniques based on geometry, texture and three‐dimensional data. Findings Most commercial iris recognition systems are based on the algorithms developed by one man, John Daugman. Whilst iris systems can be used to check a person's identity against a large database of enrolled people, face recognition systems are currently only capable of use in one‐to‐one recognition mode, or in identification mode against a very small database. The iris is very distinctive and stable over time, but the face is much more variable and therefore difficult to identify with accuracy. Originality/value Provides the general scientific reader with some insight into the specialised field of biometric recognition.