A novel approach for gesture control video games based on perceptual features: modelling, tracking and recognition

摘要

Gesture recognition has been an attractive research area for decades. Recently, the video game industry has become the major driving force for the development of advanced gesture control technologies. Conventional video games are controlled via physical devices. In contrast, the emerging trend is using camera-based human computer interface (HCI) to capture human gestures and control game playing directly.rnThis paper presents a novel approach for facilitating the development of gesture control-based video games. A time-of-flight (TOF) camera is adopted to provide both depth and greyscale image sequences. 3D perceptual gesture features are extracted and grouped into a generic gesture representation for target gesture recognition. The game control parameters are derived from the recognised gestures on the fly. This framework includes five key modules:rn1. perceptual feature extractionrn2. object tracking by perceptual groupingrn3. representation and modellingrn4. gesture recognitionrn5. game control parameter generation.rnA proof-of-concept dart game is implemented for demonstration and evaluation.