TNA project : Application of hand gestures in VR environment using 5DT 14 Ultra data glove

Acronym : 77-Application of hand gestures in VR environment using 5DT 14 Ultra data glove-Luanin

Project Lead : Ognjan Luanin From : University of Novi Sad, Faculty of Technical Sciences, Dept. of Production Engng.

Dates : from 14th January 2014 to 21st January 2014

Description :



Motivation and objectives :
Analysis and recognition of hand gestures has been focused by the research community during the last two decades. Such attention is motivated by the importance of practical application of hand gestures, primarily in the field of virtual reality, sign language recognition, medical rehabilitation, and computer animation. The complexity of hand gesture recognition problem has resulted in a number of different approaches. Among the solutions which have been established as standard, there is a glove-based approach which uses specially designed gloves with sensors of variable type and number to register finger flexion or contacts between them in real time, and the vision-based approach which relies on various types of image capture equipment. The main objective of this research is to enhance glove-based gesture recognition through application of artificial intelligence and data mining methods. Another goal is to enhance the standard gesture dictionary of 5DT data gloves with complex static gestures, which allow each finger to be flexed at an arbitrary angle. These gestures are more ergonomic than the joint-limit gestures, used by the 5DT data glove family, whose configuration is based on either fully open or fully flexed fingers.

Teams :
As part of the Faculty of Technical Sciences (University of Novi Sad), the Department for Production Engineering consists of 4 Chairs and 12 Laboratories, and has so far realized more than 200 domestic and international research projects. Currently, the Department is involved in 10 international scientific projects, from bilateral to FP7.

Dates :
starting date : 14 January, 2014
ending date : 21 January, 2014

Facilities descriptions :
http://visionair-browser.g-scop.grenoble-inp.fr/visionair/Browser/Catalogs/VDS.PL.html

Recordings & Results :
Participation in the VISIONAIR project allowed to test the scallability of the Probabilistic Neural Network - PNN-based gesture recognition system, originally designed for the 5-sensor data glove.

Conclusions :
As the result of the first stage of this investigation, during the gesture acquisition process, each of the five subjects produced 12 data files, in total – 60 files which need to be processed. Principal component analysis performed in the subsequent step of the processing, generated the PCA scores and it was possible to see that, for all of the subjects, all gestures could be efficientlz reduced to vectors consisting of just several members which represent the components which account for more than 90% of the variance.




Few images :