Title (eng): Development and Evaluation of a New Palm Vein Recognition System
Fischer , M. (Matthias)
St. Pölten, FH-Stg. Information Security, Dipl.-Arb., 2011
The purpose of this master thesis is to develop a palm vein recognition system based on the Enhanced Local Gabor Binary Patterns Histogram Sequence Algorithm used for feature extraction and matching. Since this approach is originally and successfully used in face recognition systems with a recognition rate up to 99% its application on vein patterns is analyzed. Since in vein recognition systems different vein pattern types can be adducted, they are characterized and compared considering the general biometric requirements. Subsequently the palm vein pattern type as the most preferred type is adopted for the development. By reason that a biometric system consists of the processing steps feature extraction and matching, in addition necessary pre-processing steps for localizing and separating the essential vein pattern information are considered and implemented. Binarization is adducted to segment the hand from the background. Afterwards the hand contour is traced and the Region of Interest - containing the vein pattern information - is localized and separated from the hand vein image for the subsequent feature extraction and matching steps. Tests are performed on the basis of a palm vein pattern database to determine the accuracy of the above approach. Therefore different test parameters are varied to obtain the ideal values for the False Rejection Rate and False Acceptance Rate. By comparing the rates with common deployed vein recognition approaches the accuracy of Enhanced Local Gabor Binary Patterns Histogram Sequence Algorithm is established. The approach achieves a FRR of 1.7% and a FAR of 0%. The rates are comparable to state-of-the-art vein recognition approaches.
Object languages: German
© All rights reserved
Classification: Biometrie; Hand; Blutgefäß