An Enhanced Backpropagation Neural Network for Multimodal Biometric System
K. Krishneswari, S. S. Arumugam · International Review on Computers and Software (IRECOS) · 2014
Multimodal biometric systems have many advantages over unimodal systems. Combining the evidence from different modalities using an effective fusion scheme can greatly improve the overall accuracy of such systems. The proposed methodology extracts DCT, Gabor and Line features from the fused palmprint and fingerprint image. The fused features are reduced using PCA and IG. The reduced feature vectors were classified using the proposed enhanced neural network classifier. The proposed method shows a significant improvement on different datasets compared to using palmprint technique alone.