Pfinder is descended from a variety of interesting experiments in human-computer interface and computer mediated communication. Initial exploration into this space of applications was by Krueger [11], who showed that even 2-D binary vision processing of the human form can be used as an interesting interface. More recently the Mandala group [1], has commercialized and improved this technology by using analog chromakey video processing to isolate colored gloves, etc., worn by users. In both cases, most of the focus is on improving the graphics interaction, with the visual input processing being at most a secondary concern. Pfinder goes well beyond these systems by providing a detailed level of analysis impossible with primitive binary vision.
Pfinder is also related to body-tracking projects like those by Rehg and Kanade [16], Rohr [17], and Gavrila and Davis [8] that use kinematic models, or those by Pentland and Horowitz [15] and Metaxas and Terzopolous [14] that use dynamic models. Such models, however, require reasonably accurate initialization, a currently unsolved problem. In addition, despite some efforts to handle occlusion, currently such models cannot reliably deal with large occlusions. Finally, such approaches require relatively massive computational resources to run in real-time.
Pfinder is perhaps most closely related to the work of Bichsel [5] and Baumberg and Hogg [4]. These systems segmented the person from the background in real time using only a standard workstation. Their limitation is that they did not analyze the person's shape or internal features, but only the silhouette of the person. Consequently, they cannot track head and hands, recognize any but the simplest gestures, or determine body pose.