Performance Evaluation of Multimodal Biometrics Using Particle Swarm Optimization

In any multimodal biometric system, an effective fusion method is necessary for combining information from various single modality systems. This requires that the scores from different modalities be in the same range so that they can be combined properly. Hence normalization techniques are needed for efficient fusion of scores. Performance of various normalization and fusion techniques are examined by using the five modalities namely iris, palm print, retina, ear and fingerprint. Particle swarm optimization technique is then applied on different fusion methods to optimize the fusion parameters such as threshold and weights and it shows improved recognition rates compared to the fusion of original scores. The results are experimentally verified and the performance graphs are plotted and better fusion method is recognized.