Representation of dorsal hand vein pattern using local binary patterns (LBP)

Abstract

In this revolutionized and digital world, the increasing need of security to protect individuals and information has led to a rise in developing biometric systems over traditional security systems such as pincode and password. Finding more reliable, practical and more acceptable biometrics and techniques are attracting the attention of researchers. Recently, hand vein pattern biometrics has gained increasing interest from both research communities and industries. Researchers are exploiting the different biometric phases by applying existing techniques or devising new ones to develop enhanced biometric systems. Up to now, most researchers have thinned the dorsal hand vein pattern and apply corresponding techniques for feature representation and matching. However, not many techniques have been explored with relation to considering the whole hand vein image. In this research work, local binary pattern, which is a powerful technique for representing texture description of an image, have been applied on dorsal hand vein images. This method outperforms existing vein representation techniques by having a recognition rate of 98.4% on a database of more than 1000 images. In addition, this proposed method has no effect on rotated images, which is desirable in any biometric security system.