Chung-Yi Lin, Sheng-Wen Shih and Yi-Ping Hung
In this paper, we propose a new automatic approach to reconstructing a model for the 3D environment by use of an active binocular head. To efficiently store and access the depth estimates, we propose the use of the inverse polar octree which can transform both the unbounded estimate and the unbounded estimation error into a bounded 3D space with appropriate resolution. The depth estimates are computed by using the asymptotic Bayesian estimation method, which includes the use of Markov random fields. In order to apply this method, the active binocular head (the IIS head) has been calibrated with very high accuracy. The path of the local motion required by the asymptotic Bayesian method is determined online automatically to reduce the ambiguity of stereo matching. Some rules for checking the consistency between the new observation and the previous observations have been developed to properly update the inverse polar octree. Experimental results have shown that the proposed approach is very promising for automatic generation of 3D models which can be used for rendering a 3D scene in a virtual reality system.