Multibiometric system design

In this thesis, e-passport system compliant biometric system modelling has been achieved by using fingerprint and face multibiometrics. Feature extraction for fingerprint data has been done by using a new method, Angle Invariant Fingerprint Matching (AIFM). The method proposed for face recognition has been developed by considering International Civil Aviation Organization (ICAO) standards compatible passport pictures. This method namely Relational Bit Operator (RBO) is used during feature extraction. While template has been being extracted via RBO from face pictures, a parallel algorithm named as slice, process, merge similar to divide and conquer approach is run on the multicore architecture by using load balancing. Biometric templates are cryptographically embedded into the QR codes. Encryption key is hidden into the QR image via steganography. Compressed QR is transferred into the matcher module in a secure way for decision level fusion of fingerprint and face templates. Thesis has been completed with analysis and tests by considering logic level hardware design appropriateness of every proposed method in each step for future studies.