Main Article Content
Biometrics refers to the automatic authentication of a living person based on physiological or behavioral characteristics. Hand recognition involves an analysis and measures of the features of the hand. In this work, an authentication system based on hand geometry (i.e., fingers' features) using artificial neural network are proposed and implemented. The features of the fingers are extracted from the hand images which are producing after passing through a sequence of preprocessing stages. The authentication process consists of two phases, enrollment phase: image capture, smoothing, binarization, gaps removal, edge detection, thinning and chain coding were implemented. In the verification phase, feature vector to the unknown person is extracted from its hand image and determine the most discriminating features. The proposed system suggested the BPNN for training and testing data. In final step, we encode and decode features to produce hash value based on base64 method. In this research the used data is the CASIA dataset, which consists of several image snapshots taken by scanners working at different wavelengths, in this study all the snapshots belong to different wavelengths are taken. The experimental results showed that the extracted feature vector has good discrimination capabilities that led to a recognition rate over 99.50%.