Investigation of Task based Gesture Recognition for Laser Actuated Presentation Systems

Atul Chowdhary, Subhajit Karmakar, Vivek Agrawal, and Sandip Sarkar

Keywords

Hand movement, Laser controlled mouse, Human-computer interaction, Gesture detection

Abstract

We revisit the idea that human subjects could control an on-screen presentation using pointing and drawing gestures made using a laser pointer on a presentation screen. One of the major hurdles of this kind of systems is to unambiguously distinguish between voluntary and involuntary gestures. We have investigated in this work specific spatio-temporal signature of gesture movements that could be used to distinguish between voluntary and involuntary gestures. Three simple gestures that could be made using a laser pointer; uni-directional drawing gesture, bi-directional drawing gesture and pointing gesture, had been used for the study. Subjects were asked to perform these tasks, with a laser pointer, on a screen. Our study showed that though the subjects tended to perform the tasks at their preferred tempo and the velocity profile of all subjects exhibited a task specific characteristic feature. Similar study with pointing gestures also revealed that the position of the laser pointer exhibited anisotropic Gaussian distribution, whose standard deviations decreased with the increase in dwell time.

Important Links:

Go Back