Fusion operators for multi-modal biometric authentication based on physiological signals

The most basic operators, like the sum and the product, have been used for data fusion in many application fields together with ordinal operators and the majority voting operator since the early stages of research. These application fields include biometrics, which constitutes the focus of the paper presented herein. All these operators have evolved into more advanced ones, particularly through the results of soft-computing and fuzzy operator research. However, these advances in state of the art have not been transfered to the different application fields. The presented work provides a comparison of different soft data fusion operators in a biometric application. Hence we analyze the performance of their application in a multimodal system, which takes into account two modalities based on physiological signals, electroencephalogram (EEG) and electrocardiogram (ECG). The analysis is done by evaluating the performance of five operators on a 29 subject database. The performance improvement due to the application of a soft data fusion stage is evaluated and demonstrated.