Presented by: Rong Wang

Abstruct:

Predicting performance of biometrics is an important problem in a real world application. In this paper we present a binomial model to predict fingerprint recognition performance. We use a fingerprint identification algorithm to find the number of corresponding triangles as the match and non-match similarity scores. Then we use these scores in a binomial prediction model, which uses small gallery to predict performance on a large population. The results on the entire NIST-4 database show that our model can reasonably predict large population performance.