Knowledge extraction from human motion

Observing people is currently one of the most active application areas in computer vision. This strong interest is driven by a wide spectrum of promising applications in many areas such as virtual reality, smart surveillance, perceptual interface, etc [13]. This paper presents the concept of knowledge extraction from single human motion via a fixed camera in an enclosed environment in order to mine some movement attributes. We propose a framework based on five mining tools. The five measurements are extracting pixel coverage of a particular object, time domain, frequency distribution of pixels of interest, distances crossed in each frame and considering the object velocity. We assume that, taking into account the measurements mentioned above will introduce a robust knowledge extraction approach.