Multibiometrics enhancement using quality measurement in score level fusion

Abstract

Multi-biometrics is a solution which is sought to overcome functional and security deficiency in a baseline biometric configuration. In this paper, we propose a multi-biometrics scheme and we apply the cross validation between two databases to study the Equal Error Rate improvement of score level fusion. Our fusion function is constructed using an evolutionary GA on the XM2VTS score database. The best one is tested on a sub-sequence of the BioSecure Score database. As this database offers quality measurement, we transform our function into a weighted function with user-specific approach to study performance enhancement with quality integration. The results are significantly improved with a high confidence and quality measurement becomes inherent to reduce recognition errors.