Repetition-based Dense Single-View Reconstruction

This paper presents a novel approach for dense reconstruction from a single-view of a repetitive scene structure. Given an image and its detected repetition regions, we model the shape recovery as the dense pixel correspondences within a single image. Correspondence is represented by an interval map that tells the distances of each pixel to its matched pixels with in the single image. In order to obtain dense repetitive structures, a new repetition constraint is developed, which penalizes the inconsistency between the repetition intervals of the dynamically corresponding pixel pairs. We deploy a graph-cut to balance between the constraint of high-level geometric repetition, the photometric consistency and the spatial smoothness of the reconstructed scene. We demonstrate the accurate recovery of interval maps and dense 3D repetitive structures through our proposed method on a variety of experiments. Furthermore, these experiments prove the robustness to outliers such as structure varia...

• Type: Journal
• Year: 2011
• Where: CVPR
• Authors: Changchang Wu, Jan-Michael Frahm, Marc Pollefeys
• Keywords: Computer Vision, CVPR 2011, Interval Maps, Pixel, Repetitive Structures