Real-Time Issues in On-Chip Hand Gesture Recognition

摘要

An on-chip hand gesture recognition system provides an efficient means to integrate a natural, unobtrusive interface into a diverse set of everyday devices. Although many systems emphasize high-speed performance, they are unable to meet critical real-time demands, such as deadlines and resource limitations. This paper addresses the hardware and software issues that are deemed critical for successful real-time gesture recognition. Formal techniques including Petri-nets are used to develop a model of the system which is then used to optimize both hardware and software system components. It is proposed that use of a dynamically reconfigurable architecture is necessary to efficiently implement the diverse but periodic tasks of hand gesture recognition on-chip.