k-Nearest-Neighbour based Numerical Hand Posture Recognition using a Smart Textile Glove
2015 (English)In: AMBIENT 2015: The Fifth International Conference on Ambient Computing, Applications, Services and Technologies / [ed] MaartenWeyn, International Academy, Research and Industry Association (IARIA), 2015, 36-41 p.Conference paper (Refereed)
In this article, the authors present an interdisciplinary project that illustrates the potential and challenges in dealing with electronic textiles as sensing devices. An interactive system consisting of a knitted sensor glove and electronic circuit and a numeric hand posture recognition algorithm based on k-nearestneighbour (kNN) is introduced. The design of the sensor glove itself is described, considering two sensitive fiber materials – piezoresistive and piezoelectric fibers – and the construction using an industrial knitting machine as well as the electronic setup is sketched out. Based on the characteristics of the textile sensors, a kNN technique based on a condensed dataset has been chosen to recognize hand postures indicating numbers from one to five from the sensor data. The authors describe two types of data condensation techniques (Reduced Nearest Neighbours and Fast Condensed Nearest Neighbours) in order to improve the data quality used by kNN, which are compared in terms of run time, condensation rate and recognition accuracy. Finally, the article gives an outlook on potential application scenarios for sensor gloves in pervasive computing.
Place, publisher, year, edition, pages
International Academy, Research and Industry Association (IARIA), 2015. 36-41 p.
Computer Science Textile, Rubber and Polymeric Materials
IdentifiersURN: urn:nbn:se:his:diva-12105ISBN: 978-1-61208-421-3OAI: oai:DiVA.org:his-12105DiVA: diva2:917517
AMBIENT 2015: The Fifth International Conference on Ambient Computing, Applications, Services and Technologies, July 19 - 24, 2015, Nice, France
Research supported by Västra Götalandsregionen (VGR), grant number RUN 612-0197-13.2016-04-062016-04-062016-08-18Bibliographically approved