Vision-based tracking with dynamic structured light for video see-through augmented reality

摘要: Tracking has proven a difficult problem to solve accurately without limiting the user or application. Vision-based systems have shown promise, but are limited by occlusion of landmarks. We introduce new approach vision-based tracking using structured light generate The novel aspect this is system need not know 3D locations This implies that motion within field view camera does disturb as long landmarks reflected off any surface into camera. This dissertation specifies an algorithm which tracks light. A simulator demonstrates excellent performance on data from application currently inaccurate tracking. Further analysis reveals directions for implementation system, theoretical limitations, and potential extensions algorithm. The term augmented reality (AR) been given applications merge computer graphics with images user's surroundings. AR could give doctor “X-ray vision” examine patient before during surgery. At point in time, used place traditional methods performing medical other tasks. One important limits acceptance lack precise registration—alignment—between real synthetic objects. There many components contribute registration. One most system. must be accurate, so objects aligned properly. Our work focuses applications. These require alignment imagery physician's patient. Although technologies applied—mechanical, magnetic, optical, et al—we yet find sufficiently accurate robust provide correct reliable We believe specified here contributes two key ways: it takes advantage equipment already AR, sufficient registration demanding imposing limitations current systems.