Multi-modal hand-based biometric system using energy compaction of various transforms and wavelets

The need for stringent security and accurate identification is to a large extent accomplished by multi-modal biometric systems. These systems counter-verify the identity using more than one biometrics. In this paper we have used three hand-based biometric viz. fingerprint, palmprint and finger-knuckle print. The main advantage of this system is ease of acquiring biometric and hence high user's acceptability. Individual transforms like DCT, Walsh, Kekre and Haar and their combinations resulting in hybrid transforms, hybrid two-resolution wavelets and hybrid multi-resolution wavelets have been applied to each biometric. Energy compaction of these transforms and wavelets has been used to generate feature vectors. Both open set and closed set experiments have been performed on partial fingerprint database and palmprint and finger-knuckle print databases from Hong Kong polyu and plots of efficiency of closed set and receiver operating characteristics of open set have been generated. With 100% efficiency for hybrid multi-resolution wavelet with combination of Walsh-Kekre and 0% Equal Error Rate (EER) and 100% Security Parameter Index (SPI) for hybrid multi-resolution wavelet with combination of Walsh-DCT and Haar-DCT, this method is best suitable for applications with relatively smaller database.