Search for a command to run...
This book presents a novel method of multimodal biometric fusion using a random selection of biometrics, which covers a new method of feature extraction, a new framework of sensor-level and feature-level fusion. Most of the biometric systems presently use unimodal systems, which have several limitations. Multimodal systems can increase the matching accuracy of a recognition system. This monograph shows how the problems of unimodal systems can be dealt with efficiently, and focuses on multimodal biometric identification and sensor-level, feature-level fusion. It discusses fusion in biometric systems to improve performance. • Presents a random selection of biometrics to ensure that the system is interacting with a live user. • Offers a compilation of all techniques used for unimodal as well as multimodal biometric identification systems, elaborated with required justification and interpretation with case studies, suitable figures, tables, graphs, and so on. • Shows that for feature-level fusion using contourlet transform features with LDA for dimension reduction attains more accuracy compared to that of block variance features. • Includes contribution in feature extraction and pattern recognition for an increase in the accuracy of the system. • Explains contourlet transform as the best modality-specific feature extraction algorithms for fingerprint, face, and palmprint. This book is for researchers, scholars, and students of Computer Science, Information Technology, Electronics and Electrical Engineering, Mechanical Engineering, and people working on biometric applications.