查看文章

Usage of human physical and behavioural traits for individual identification dates back to the beginning of human civilization. The demand for automation of biometric system was raised as a result of increase in population. Nowadays, in many applications biometric authentication is gaining popularity. However, its performance is still a constraint for its wide scale implementation. In our proposed work, fingerprint and ear are considered as it provides higher performance than unimodal biometric. Our objective is to store the biometric in the database as fuzzy vault, as storing the biometric template directly in to the database weakens the system security. Initially, the input images of fingerprint and ear are pre-processed to filter the noise and to improve the image contrast. From the pre-processed fingerprint, minutiae points are extracted. Upon the pre-processed ear image, the dimension reducing technique, Principal Component Analysis (PCA), is applied to extract global features. Texture feature is also extracted from both the modalities using Local Gabor XOR Patterns (LGXP). To these extracted features cuckoo optimization is applied to generate optimized points. Chaff points are generated, from randomly generated secret keys. These chaff points are fused with optimized points to generate fuzzy vault. The sensitivity, specificity and accuracy are calculated for the cuckoo optimized fuzzy vault authentication system and compared with the non-optimized fuzzy vault system. Copyright©