BIOMETRIC FUSION BASED ON IRIS & THUMBPRINT USING ARTIFICIAL NEURAL NETWORK WITH OPTIMIZATION

In the authentic world applications, biometric systems often face limitations because of feature selection, noise, size of data etc. Multi biometric systems are used to overcome this type of difficulty by providing various pieces of confirmation of the same identity. The proposed work belongs to biometric safekeeping domain for the identification and authentication. It gives explanation to the predicament of identification with lower errors, high accuracy, and less complexity of the proposed system. For solving the problem in proposed work, a multimodal biometric system by combining iris and thumb impression at match score level using artificial neural network (ANN) is being developed. In the proposed work, we present the biometrics recognition system based on the iris and thumbprint using the artificial neural network as a classifier. There are two sections, first section is iris recognition and second section is thumbprint recognition, after that we fused the result of both system to achieve better results. The work is being designed and developed on the basis of ANN as a classifier and as a feature extraction technique we use SIFT and minutia for iris and thumbprint respectively with BFO (Bacterial foraging optimization) and GA (Genetic Algorithm) as optimization techniques. Our investigational results suggest that the ANN method for the recognition at the decision level is the most excellent followed by the different techniques like Sum Rule, SVM, Clustering and KNN. The performance evaluation of proposed technique is reported in terms of FAR, FRR, and Accuracy after doing comprehensive tests on the CASIA-Iris databases for iris and the FVC 2004 fingerprint database and we concluded the accuracy of proposed system is more than 98% with a better FAR and FRR value..