Implementation and Evaluation of SIFT Descriptors based Finger-Knuckle-Print Authentication System

Abstract

Finger-Knuckle-Print (FKP) uses feature detection and matching techniques in its hard core design. It works similar for almost every authentication system. The Scale Invariant Feature Transform (SIFT) is the most reliable feature extraction technique that is used in authentication systems on FKP. The feature descriptors detected by SIFT claim to be capable of distinguishing each and every image in the dataset from one another with the cost involved in its operations. In SIFT based FKP authentication systems, the storage and computational cost will directly depend on the size of the feature descriptors used. Such matching process will directly match these feature descriptors to find an exact match and the descriptors were directly stored in storage media as templates. Hence there is a necessity for storing all the feature descriptors of the enrolled FKP images for future references. The size of these feature descriptors data will be greater than the original FKP image dataset and the performance of the system will rapidly decrease with respect to the increase in enrollment in the database. The proposed work address these issues with FKP based authentication system using SIFT for efficient computation and cost compared with the existing work and proven to be secure and tough resistant for authentication system.