Abstract: Biometric-based systems are used for human recognition in real-time applications. Due to the huge size of biometric database in current methodologies, more time is required for image search in the database. In this study, an innovative technique is introduced called Two-Modal Biometrics based Content-Based Image Retrieval approach (TMB-CBIR) for biometric authentication. In the two-modal biometrics, iris and fingerprint images are considered. In the iris image, both color and texture features are extracted where color feature is used for indexing and text feature is used to find the similarity of images extracted by using Speeded Up Robust Features (SURF) algorithm. In the fingerprint images, the Improved Locality-Sensitive Hashing (ILSH) indexing method is used that considers the locality of points so that the nearby points remain closer instead of recognize perfect match. In this method, the data is disseminated and every data point is uniformly hashed. Union of Candidate Lists fusion method is used to merge both lists of candidate identities output. Finally, the set of reference images are chosen from convex hull of feature space in order to reduce the estimation time. Experimental results show that TMB-CBIR achieves high accuracy and less response time.
D. Binu and P. Malathi, 2016. Two-Modal Biometrics Based Content-Based Image Retrieval Approach (TMB-CBIR) for Biometric Authentication. Asian Journal of Information Technology, 15: 2154-2161.