Estimating Planar Patches for Light Field Reconstruction

Mullins, Andrew; Bowen, Adam; Wilson, Roland; Rajpoot, Nasir M.

doi:10.5244/c.19.23

Cited by 2 publications

(1 citation statement)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These estimates are used as a prior in a multiscale particle filter, which provides the final surface estimates, in terms of a number of disjoint quadrilaterals, corresponding to square blocks in each image of the scene. The patches can then be tracked over time using a second particle filter [14], and used to reconstruct arbitrary views of the scene using an adaptation of a conventional graphics renderer, giving real-time reconstructions [13], [17].…”

Section: Surface Estimation Methodsmentioning

confidence: 99%

Bayesian Surface Estimation from Multiple Cameras Using a Prior Based on the Visual Hull and its Application to Image Based Rendering

Mullins

Bowen

Wilson

et al. 2007

Procedings of the British Machine Vision Conference 2007

Self Cite

View full text Add to dashboard Cite

The problem of visible surface estimation for image-based rendering is tackled using a new approach, which combines visual hull and surface estimation techniques. It is shown that the new method combines the best features of both approaches, being more robust than direct surface element estimation and more flexible than the visual hull. The new method uses an estimate of the visual hull as a prior on the ill-posed problem of surface element estimation. To improve the computational preformance of the algorithm, a multiresolution approach to surface patch estimation is used. The patches thus estimated can then be tracked over time, to provide an accurate model of the surface geometry, which is then used for estimation of the scene from arbitrary viewpoints. After a brief description of the algorithm, results are presented to show the improvement in performance which can be obtained using either technique alone. The paper concludes with a discussion of extensions to the work currently under investigation.

show abstract

Section: Surface Estimation Methodsmentioning

confidence: 99%