Abstract

Biometric as a method for recognizing individual identity based on physical or behavioural traits has been one of interested research areas. Since biometrical verification based on hand geometry shows some advantages, it has been more implemented than other biometrics. The advantages include easy to use, easy to collect data, high public acceptance, and efficient for data storage. Biometrics based on hand geometry also has limitation on uniqueness so it is ideally suited for verification (authentication) rather than identification. In this paper we designed a system which grabbed hand images and extracted some features including width of fingers, length of fingers, and width of palm. Feature extraction was obtained from estimating parameter of likelihood equalization. The metrics used for calculating distance value were absolute and Euclidean metric. This system was applied to 130 hand images that acquired in first session and 180 hand images that acquired in second session. Experimental results indicated overall acuracy, FAR, and FRR, respectively were 98,179%, 0,216%, 1,605%. On the other hand, this system has limitition in the uniqueness. Some individuals may have similar nature of hand geometry. Therefore, instead of idetification purpose, this system is often used for verification or authentication purpose [2]. 2 SYSTEM MODEL Figure 1 shows system diagram which contains two main phases including enrollment and verification. Both phases involve steps of image image acquisition, preprocessing, feature extraction, database feature, matching or comparison, and decision.

Citations