Fusion in Multimodal Biometric System: A Review

Objectives: The use of multimodal biometric has been introduced recently owing to use of multiple biometric modalities. Here we perform in-depth review of the various methods used for multimodal biometric technology. Methods/ Statistical Analysis: Here we present a systematic review of various methods used for fusing multiple biometric modalites. Specifically, fusing at various levels such as, before matching and after matching. Score level, feature level, rank level and decision fusion is followed by feature optimization using methods such as genetic algorithms and artificial neural networks. Findings: Single biometric based methods suffer from lack of security and efficiency. This leads to advent of multimodal biometric systems. However, fusing various biometric modalities is being persued with very high interest. We describe the granular nature of several methods used to fuse multiple biometric modalities. A wide range of methods are being employed to fuse biometric data. These methods vary in efficiency and are highly dependant upon the selection of type of biometric chosen for fusion. Application/Improvements: As computational efficiency increases, there increase in more secure and efficient biometric systems that use multiple sources of biometric identification and access authorization.