FUZZY VAULT FUSION BASED MULTIMODAL BIOMETRIC HUMAN RECOGNITION SYSTEM WITH FINGERPRINT AND EAR
Amitabh Wahi · 2014
Human Recognition is one of the admired tasks over the world for recognizing a person using biometrics by determining physical or behavioral characteristics of that person. In our existing work, we have alrea dy worked out a multimodal biometric recognition syste m with fingerprint, palm print and hand vein. For getting more accurate recognition of our biometric system, in this work, we use ear as one of the moda lities with the fingerprint. In order to improve the clear visible of input image databases, pre-processing o f images is initially done. After the pre-processing of these images only, the features from the fingerp rint and ear modalities are extracted clearly for the furthe r processes. In the fingerprint images, the minutia e features are extracted directly and from the ear, the shape features are extracted using Active Appearance Model (AAM). Then, a grouped feature vector point is gain ed using chaff points and these two extracted featu re points. After acquiring the grouped feature vector points, the secret key points are attached with the grouped feature vector points to formulate the fuzz y vault. Finally, test person’s grouped vector is m atched up to the fuzzy vault data base to the accurate rec ognition of the correct person. Our proposed work i s effectively evaluated in Matlab with the evaluation metrics FAR, FFR, GAR and Accuracy by changing the secret key size at every time. The results of our p roposed work facilitate very better values for the recognition of persons with the fingerprint and ear modalities. Moreover, our existing work is also compared with our proposed work for proving that our proposed work is good. In addition to this, other existing work papers are also taken for our compari son work, which clearly proves that our proposed wo rk outperforms other techniques by providing very much better recognition accuracy.