— We propose a novel inertial-aided KLT feature tracking method robust to camera ego-motions. The conventional KLT uses images only and its working condition is inherently limite...
Modern structure from motion techniques are capable of building city-scale 3D reconstructions from large image collections, but have mostly ignored the problem of largescale struc...
This contribution addresses the problem of obtaining photorealistic 3D models of a scene from images alone with a structure-from-motion approach. The 3D scene is observed from mul...
It’s common experience for human vision to perceive full 3D shape and scene from a single 2D image with the occluded parts “filled-in” by prior visual knowledge. In this pa...
Abstract. We present some recent progress in designing and implementing two interactive image-based 3D modeling systems. The first system constructs 3D models from a collection of ...
Heung-Yeung Shum, Richard Szeliski, Simon Baker, M...