J Starck

Dual-Mode Deformable Models for Free-Viewpoint Video of Sports Events

Generating free-viewpoint video in outdoor sports environments is currently an unsolved problem due to difficulties in obtaining accurate background segmentation and camera calibration. This paper introduces a technique for the reconstruction of a scene in the presence of these errors. We tackle the issues of reconstruction completeness, and accuracy of surface shape and appearance. We introduce the concept of the conservative visual hull as a technique to improve reconstruction completeness. We then present a view-dependent surface optimisation technique using deformable models to improve surface shape and appearance. We contribute a novel dual-mode snake algorithm that is robust to noise and demonstrates reduced dependence on parameterisation by separating the search of the solution space from the data fitting. We conclude by presenting results of this technique along with a quantitative evaluation against other reconstruction techniques using a leave-one- out data set.

Efficient Dense Reconstruction from Video

We present a framework for efficient reconstruction of dense scene structure from video. Sequential structure-from-motion recovers camera information from video, providing only sparse 3D points. We build a dense 3D point cloud by performing full-frame tracking and depth estimation across sequences. First, we present a novel algorithm for sequential frame selection to extract a set of key frames with sufficient parallax for accurate depth reconstruction. Second, we introduce a technique for efficient reconstruction using dense tracking with geometrically correct optimisation of depth and orientation. Key frame selection is also performed in optimisation to provide accurate depth reconstruction for different scene elements. We test our work on benchmark footage and scenes containing local non-rigid motion, foreground clutter and occlusions to show comparable performance to state of the art techniques. We also show a substantial increase in speed on real world footage compared to existing methods, when they succeed, and successful reconstructions when they fail.

Transfer of albedo and local depth variation to photo-textures

Acquisition of displacement and albedo maps for full building façades is a difficult problem and traditionally achieved through a labor intensive artistic process. In this paper, we present a material appearance transfer method, Transfer by Analogy, designed to infer surface detail and diffuse reflectance for textured surfaces like the present in building façades. We begin by acquiring small exemplars (displacement and albedo maps), in accessible areas, where capture conditions can be controlled. We then transfer these properties to a complete photo-texture constructed from reference images and captured under diffuse daylight illumination. Our approach allows super-resolution inference of albedo and displacement from information in the photo-texture. When transferring appearance from multiple exemplars to façades containing multiple materials, our approach also sidesteps the need for segmentation. We show how we use these methods to create relightable models with a high degree of texture detail, reproducing the visually rich self-shadowing effects that would normally be difficult to capture using just simple consumer equipment.