hi

In this experiment, we use two or more fingers types of the same person to increase the performances of our system. Thus, we can enhance the recognition performance of the FKP recognition algorithm by using multiple fingers. After the generation of the LFVs for different finger types of the same user (LFV for LMF, LFV for LIF, LFV for RMF and LFV for RIF), it is possible to fusion these two or more vectors in a single vector for each user. We use 6 vectors for training for each person (990 vectors for the training) and 6 vectors for test (990 vectors for test). These vectors of all users are then subjected to subspace projection LDA. The nearest neighbor classifier and cosine similarity distance are used for respectively matching and decision stages. Table.2 tabulates the ROR (%) for various combinations possible. Thus, the proposed system multi-modal gives the ROR=98.69% for the combinations (RIF-LMF) and (LMF-LIF). ROR was 98.59% for the combination (RMF-LIF). It achieves the RORs 98.08% and 98.28% for the combinations (RIF-LIF) and (RIF-RMF), respectively. As can be seen, the best result was obtained with the combination of three or four fingers. It can achieve even higher precision, an ROR of 99.90% for the combination (LMF-LIF-RMF). ROR was 99.80% for the combination (RMF-RIF-LMF) and was 99.49% for the combination (RMF-RIF-LIF). Finally, ROR was 100.00% for the combination (LMF-LIF-RIF) or all fingers. Comparisons in terms of the ROR (%) between our method and the other two methods in [10, 15] under the same condition are shown in Figure 10. It is clear that the proposed system gives the highest ROR.