Visual Human-Computer Interaction

Technologies such as Virtual and Augmented Reality has gained extensive popularity in the recent years. Simultaneously, vision systems and computing power have reached a point, where it is possible to acquire and process geometric and appearance data to produce photorealistic renderings that can appear indistinguishable from real photographs. This enables new ways for Human-Computer Interaction (HCI) methods and applications, which should be further evaluated to explore their full potential. This thesis addresses a set of vision based challenges concerning HCI. The presented contributions fall into the overall themes of geometric acquisition of refractive objects, photorealistic rendering for computer graphics applications, and demonstration of systems for advanced and realistic applications for HCI. Accordingly, the work of this thesis is presented in a four-element taxonomy: Geometry and appearance digitization, tracking, visualization and interaction, and datasets. The work contributes to state of the art methods and prepares the ground for future research within the above-mentioned topics and generally the field of visual HCI.