Integrating prior knowledge and structure from motion

A new approach for formulating prior knowledge in structure from motion is presented, where the structure is viewed as a 3D stochastic variable, hereby priors are more naturally expressed. It is demonstrated that this formulation is efficient for regularizing structure reconstruction via prior knowledge. Specifically algorithms for imposing priors in the proposed formulation are presented.

[1]  Takeo Kanade,et al.  Image-consistent surface triangulation , 2000, Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662).

[2]  Michael Isard,et al.  Active Contours , 2000, Springer London.

[3]  Timothy F. Cootes,et al.  Active Shape Models-Their Training and Application , 1995, Comput. Vis. Image Underst..

[4]  Roberto Cipolla,et al.  Layer Extraction with a Bayesian Model of Shapes , 2000, ECCV.

[5]  Richard Szeliski,et al.  A layered approach to stereo reconstruction , 1998, Proceedings. 1998 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No.98CB36231).