Development and implementation of multibiometric on palm finger joint and palm pattern

A multibiometric approach is presented for recognition based on long vector structure with using palm finger joint surface and palm print which are from the same hand image in this paper. In this approach, initially image normalization and Discrete Wavelet Transform is applied as a preprocessing to palm finger joint surface and palm print. Afterwards these two biometrics are joined to be long vector. By using Kernel Fisher Discriminant method most discriminative features of long vector which is obtained by joinder are extracted. Support Vector Machine is utilized for classification. Finally, proposed approach is tested on 1614 palm finger joint surface and palm print images which belong to 132 different person. Identification performance for palm print is %99.87, for palm finger joint surface is %99.02 and for multibiometric structure based on long vector is %99.87. Obtained results (%99.87 accuracy rate for 132 persons) demonstrate proposed approach's success.