Feature Level Fusion Of Palm Print And Palm Vein

Biometric systems have become a major part of research due its application of identification. Code provides a multimodal biometric system using palm prints modality combined with palm print modality. DCT transformation is applied initially into input image. The proposed methodology uses standard deviation of pre-defined block of DCT coefficient as feature vector. In this way single image is converted into feature vector of 1 x 39. Recognition process is being done by performing distance measurement between feature vector of testing and training data set. Results show that the False Acceptance Rate (FAR) of feature level fusion is less than that of uni-modal systems, hence having multimodality is advantageous. Testing and training is done on database of 500 students of College of Engineering Pune, Pune, India. Canberra distance shows best result when compared to Euclidean or Manhatten distance.