Step Integration based Information Fusion for Multimodal Biometrics

The present day needs of the society for an efficient and more reliable identification system has given birth to the biometric matcher technology. Over the years, efforts have been made to improve the detection and recognition performance of the biometric systems, and at the same time reduce the processing time of the systems for a faster identification. One of the approaches for performance optimization is to fuse two or more biometric matcher technologies to add on to the performance of the individual systems. In the present paper, we propose a new step integration based fusion method for multimodal biometric technologies, based upon eliminative machine learning. The individual matcher systems are integrated in steps eliminating unwanted user classes at each matcher step. The method achieves high accuracies and recognition rates, achieving low processing times at the same time. We compare our approach with simple sum fusion technique because of the simplicity and high accuracy rates associated with the simple sum method.