Automatically localizing discontinuities during surface reconstruction

This paper describes an extension to a class of surface reconstruction algorithms based on the minimization of variation of curvature, the so-called curvature consistency framework. It deals explicitly with the problem of localizing surface discontinuities, which was one of the shortcomings of the original work. Inter-pixel interactions during the variational relaxation based smoothing process can be construed as a multiple measurement fusion problem. Using this formulation, based on "noise" properties of pixels, weights are assigned to each pixel contributing to a local surface estimate. Since discontinuities are included in this noise model, it becomes possible to obtain a smooth reconstruction of the surface from noisy data, while localizing the discontinuities.