The key objective of the paper is to summarize different process that are used in multimodal biometric system for fusion of biometric traits. The main focus is on different fusion techniques that are used for fusion having different biometric traits like palm and face, iris and fingerprint, fusion of instances of same biometric model etc. Fusion are of two types, Post Classification and Pre Classification. In Post classification method integration of information occurs after matching or classification stage. In Pre-classification information fusion, integration of information takes place before matching or prior to classification. Feature level and Sensor level fusion comes under Pre-Classification Information fusion. In this paper, we also emphasis on different normalization techniques which are very essential for fusing matching score obtained after matching process. As matching scores are of different dimensions so before taking decision normalization has to be done so that fusion will be easy.
展开▼