Intelligent Integration and Fusion of Multimodal Biometric Systems

Abstract

The propagation and the frightening expansion of frauds due essentially to the easy access to high technologies make it difficult or sometimes impossible to detect these frauds and impostures. Therefore, it has become important, if not urgent, to develop identification techniques and tools that are more robust to attacks, more precise and more efficient. There are currently in Europe and some American countries very efficient biometric systems that combine two (2) modalities (photos and fingerprints in the case of biometric passports), but they remain very vulnerable, notably because of acquisition problems, data quality or the non-permanence of certain biometrics. Therefore, the systems based on biometric modality has been growing in the world for a decade. The processes of identification and identity verification of individuals have a very wide spectrum of applications in modern society and are becoming very relevant. However, the field of multi-biometrics is not new, many researchers have been working on this topic especially in the last 10 years. Scholar Google lists nearly 6000 publications on the subject, half of which since 2018. Real systems are in production, we can cite the UID program in India to enroll the 1.300 billion inhabitants with the 10 fingerprints and the two Iris. Most of the techniques consisted in generalizing classical biometric systems (with one modality) by attribute fusion (concatenation of data, statistical reduction...), by score fusion (often performed by summing up the comparison scores from each biometric system) or by decision fusion (majority vote most often). The objective of our work is to propose new techniques of fusion & multimodal biometric integration by using artificial intelligence tools and especially the application of fusion techniques such as: Brute Force Search, Support Vector Machines, neural networks, fuzzy systems, neuro-fuzzy systems, Genetic Algorithms, Particle Swarm Optimization.

Keywords

• Biometric
• Artificial intelligence
• Pattern recognition
• Image processing

References

1. Adjoudj, R. (2006). Reconnaissance & identification of persons by automatic authentication in a high security system. (Doctoral dissertation). University of Sidi Bel-Abbès. Retrieved from http://www-inf.univ-sba.dz
2. Adjoudj, R., Gafour, A. K., Boukelif, A., & Lehireche, A. (2008). La reconnaissance des visages : une comparaison entre les réseaux des neurones compétitifs et les réseaux des neurones à spike, a comparison between artificial and spike neural network for face recognition. Revue Traitement Du Signal, 25(6), 449-458.
3. Adjoudj, R. (2022). Multi-modal biometric persons recognition system. The Eurasia Proceedings o Science, Technology, Engineering & Mathematics (EPSTEM), 21, 558-564.