Open Exhibits - Research - Papers



Methods for Multi-touch Gesture Recognition


This paper presents a number of different methods for gesture recognition. A gesture is a form of non-verbal communication in which the body performs visible actions in order to communicate a particular message. Gesture recognition, in this context, refers to the target of interpreting these gestures via mathematical algo- rithms. An analysis describing each of the algorithms advantages and disadvantages when applied to a resource-constrained mobile device in terms of recognition speed, accuracy and training time, with a certain bias towards gaming, is also presented. The most appropriate algorithm to implement in terms of these metrics is found to be Hidden Markov models. Additionally, they are easy to implement and understand. Furthermore, they require an easily managable amount of memory and disk space available on mobile devices.