On the Dynamic Selection of Biometric Fusion Algorithms

Biometric fusion involves consolidating the output of two or more biometric classifiers in order to render a decision about the identity of an individual. The authors consider the problem of designing a fusion scheme when the number of training samples is limited, thereby undermining the use of a purely density-based scheme and the likelihood ratio test statistic; the output of multiple matchers report conflicting results; and the use of a single fusion rule may not be practical due to the diversity of scenarios encountered in the probe dataset.