Y. Yasumuro, S. Yamazaki, and M. Fuyuki (Japan)
Superimposing virtual objects onto physical space, mixed reality (MR) is an effective technique for simulating an overview of a scenery including existing and synthesized objects. MR requires geometrical and photometric consistency between physical scenes and the CG (computer graphics) objects to be rendered. Marker-based approach with mirrored ball enables acquiring both geometrical and photometric (lighting) information from physical scenes through single camera input. However, the marker detec tion occasionally fails under too bright or dark lighting conditions which are desired to be reflected to the output as well. This paper proposes a method to use HDR (high dynamic range) video input for stabilizing marker recognition and capturing lighting conditions in parallel. The experimental results showed an enhanced performance in interactive rate.
Important Links:
Go Back
