A line feature approach to finger knuckle image recognition

Abstract

A vast application of biometric identification in the field of forensic purposes to identify suspect and also commonly used as time attendance system in the offices. A number of humans’ biological traits are uses for above purposes but hand-based biological traits are highly user acceptable. A lot of literature is available on fingerprint identification but it is observed that fingerprints are not always available as a source of information on the crime scene to identification suspect. Therefore, it is required to use some other hand based biological trait such as finger knuckle print (FKP) (back side of finger joint skin pattern) in order to identify the suspect. This work proposed a person identification using finger knuckle biological trait. The proposed model is empirically evaluated using Polytechnic University Hongkong FKP image benchmark dataset. The experimental results of the proposed algorithm achieved 1.8% of EER value which is very competitive over other existing algorithms and it may be used in FKP image-based person identification/ recognition systems.