Region of interest extraction for biometric cryptosystem

Abstract

Biometric technology is becoming more and more significant these days. Most of the application in store today are using biometric as a means of person authentication as it offer a convenient and easy way of authentication. Palm vein biometry is mostly used in many authentication system as it offer better security as veins are located on the subcutaneous layer of the skin and is impossible to be forged. However, there are still some issues in order to obtain high accuracy in palm vein authentication system such as some feature are not correctly extracted because of poor preprocessing process. Poor preprocessing process will produce weak keys for authentication purpose. Therefore, this paper proposed a hybrid of Gabor filter and maximum inscribe circle to obtain better region of interest for feature extraction as well as improving the accuracy of the authentication system.