Automatic Identification Based on Hand Geometry and Probabilistic Neural Networks

Recently, there has been a growing interest in biometric technology as a more reliable means for verifying or identifying persons. In this paper, we present an affordable user-friendly approach for automatic personal identification based on hand geometry and probabilistic neural networks. We evaluate and compare the performance of the proposed approach with other common classifiers including naive Bayes, rule-based, decision tree, and k-NN classifiers. The empirical results reveal that probabilistic neural networks can lead to significant improvement for user identification with more than 98% accuracy, sensitivity and specificity.