Accession Number : ADA314933
Title : Hand Gesture Recognition Using Neural Networks.
Descriptive Note : Final rept. Nov 93-Sep 95,
Corporate Author : ARMSTRONG LAB WRIGHT-PATTERSON AFB OH CREW SYSTEMS DIRECTORATE
Personal Author(s) : Morton, Paul R. ; Fix, Edward L. ; Calhoun, Gloria L.
PDF Url : ADA314933
Report Date : MAY 1996
Pagination or Media Count : 36
Abstract : Gestural interfaces have the potential of enhancing control operations in numerous applications. For Air Force systems, machine-recognition of whole-hand gestures may be useful as an alternative controller, especially when conventional controls are less accessible. The objective of this effort was to explore the utility of a neural network-based approach to the recognition of whole-hand gestures. Using a fiber-optic instrumented glove, gesture data were collected for a set of static gestures drawn from the manual alphabet used by the deaf. Two types of neural networks (multilayer perceptron and Kohonen self-organizing feature map) were explored. Both showed promise, but the perceptron model was quicker to implement and classification is inherent in the model. The high gesture recognition rates and quick network retraining times found in the present study suggest that a neural network approach to gesture recognition be further evaluated.
Descriptors : *NEURAL NETS, *MOTION, *RECOGNITION, CONTROL, AIR FORCE, NETWORKS, MAPS, MANUAL OPERATION, STATICS, HANDS, SELF ORGANIZING SYSTEMS, ALPHABETS, RETRAINING.
Subject Categories : Psychology
Distribution Statement : APPROVED FOR PUBLIC RELEASE