Hand Gestures Expressing Numbers and Its Application to Personal Identification

This paper proposes a personal identification method to express numbers by using hand gestures. A new set of hand patterns has been devised to express each number to decrease the recognition errors and for each user to easily memorize them. The normalized correlation value between a registered grayscale image and an input grayscale image is used for recognizing each hand pattern. Threshold values are used to distinguish each number expressed by each hand pattern. FAR (False Acceptance Rate) and FRR (False Rejection Rate) are defined and calculated by using the experimental data and they can be used to decide a set of threshold values to judge each number and each person at the same time. Registration and identification procedures for personal identification are shown. A prototype for personal identification is implemented and evaluated. The identification scheme using a genetic algorithm is investigated.