Performance evaluation of biometric system

Abstract

Since biometrics may used to ensure that a person accessing information is authorized to do so, interest in biometrics for information assurance has increased recently. New biometric applications are constantly being announced while at the same time new spoofing technology is being developed to defeat them. One approach to overcoming the problem of spoofing is the use of multimodal biometric fusion. Most current research is focused on overcoming the deficiencies of a single biometric trait or reducing the false acceptance rate, both without any emphasis on the false rejection rate. Multimodal biometric fusion combines measurements from different biometric traits to enhance the strengths and diminish the weaknesses of the individual measurements. In this project we considered three types of biometrics techniques are fingerprint, handscan, and voice-scan. This project examines the use of cost function to set the threshold point such that an optimization of false acceptance and false rejection rate can be achieved. Other minimum cost thresholds with different setting of FA and FR prior probabilities and costs are also shown to be better than EER in terms of total cost. The experimental results for voice-scan show that the minimum cost is better than EER in term of combination digits, also the experiments also show that by using cost function the new threshold be more accurate and by that one could be able to find new FR and FA which provide a new EER, for example the EER=5.29% for 6 digits in normal case and by using cost function the EER became 5.28%. The experimental results on the digits combination show the cost becomes less whenever the number of combination digits becomes bigger. For 2 digits combination the min-cost is 12.5 while it is 5.287 for 6 digits combination. On the hand-scan and fingerprint-scan the experimental results were perfect by the methods were used in these tasks. Hence, by considering the cost function as one way to calculate the cost for any multimodal biometric system, the different costs depending on the application, become easier to provide.