RETRACTED ARTICLE: Improved biometric authentication using surf based variational Bayesian extreme learning machine

Over the last decade, biometric authentication systems have significantly progressed and in the upcoming years, their domain specific application is likely to boost. Multimodal biometric systems have been able to alleviate some issues faced by their unimodal counterparts. Since distinct modalities are being combined and distinct algorithms are applied to improve previous research results, greater recognition accuracy and performance has been achieved. The current work proposes a multimodal biometric authentication framework using important characteristics of the face, fingerprint, finger-knuckle-print and iris, fused together to enhance biometric system’s time efficiency. The framework extracts features through Gabor filter, Zernike moments and count key-points using surf key descriptor unlike existing techniques, such as 3 biometric fusion center. Further, for feature classification, variational Bayesian extreme learning machine technique consumes 19.65 ms, whereas, our work is implemented on MATLAB 2014Ra and is validated to consume 11.49 ms.