Personal Identification and Verification by Hand Recognition

A new methodology for the person identification and verification using hand features is presented. The features are extracted from gray level hand images, which are scanned by an ordinary commercial scanner. Contrary to other bimodal biometric systems, the palmprint and hand geometry features are acquired from the same image. On their individual performances, these features are grouped into four different feature vectors. A k-NN classifier based on majority vote rule and distance-weighted rule is employed to establish four classifiers. Dempster-Shafer evidence theory is then used to combine these classifiers in case of identification. Besides, for verification step a simple majority rule was found robust for our system. Dempster-Shafer theory has proved to be much more efficient than fusion by others methods like majority vote rule and Borda count method