Comparative Study of Features Fusion Techniques

This paper presents a unique Comparative Study of Features Fusion Techniques for Iris. In 1993, Daughman developed IrisCode and it is influenced many researchers to develop different Iris Recognition System using different techniques. Iris Features are extracted using existing techniques like 2D-FFT, DWC, LBP, PCA etc. More information is extracted by fusing multiple features. Feature Fusion is a method of integrating related information extracted from a group of Training and Testing images without losing any data. In this paper, a novel technique like LR-Fusion, RL-Fusion, DU-Fusion and UD-Fusion are used for feature fusion at pixel level. The resulted features are more informative and authentication can be improved. It also provides survey about some of the various existing Feature extraction techniques and a comparative study of all Feature fusion methods. The result concludes DU-Fusion is better approach among other fusion methods. The verification and identification performance of the current model is verified on CASIA dataset.