No abstract

The paper presents a novel approach to provide a more secured remote access to informatic systems; this approach is based on biometric identification multimodal methods with more levels of biometric fusion. The multi-level fusion is the novelty of this solution, as the actual approaches in multimodal biometric as relying on single-level fusion schemes. The integration of more fusion schemes within the same biometric system enhances performance, security and accuracy for the actual unimodal biometric systems and also for the multimodal ones, especially for on score-level fusion.

The paper presents an optimized multimodal biometric system for identification applications. This solution is based on innovative and computational-efficient biometric data hierarchical classifiers (with detection and discrimination stages) and also with a post-classification fusion. The detection-based classification is very suitable for applications with several security levels in which the end-users have various authorization degrees. The proposed solution supports an optimal trade-off between the identification accuracy and the computational complexity, which is important for the medium-and large-scale identification biometric applications.

No abstract

No abstract