Low cost RGB-D sensors allow easy access to depth data. This data is often prone to noise, missing fine details and does not have the same resolution as the RGB data. Photometric techniques however can recover fine geometric details based on an RGB image, but introduce some man-made regularity assumptions to the scenario of interest. Combining photometric techniques with RGB-D data shows that some depth prior information leads to a much more natural sensor-made regularization.

This presentation is part of Minisymposium “MS57 - Recent Trends in Photometric 3D-reconstruction (2 parts)”
organized by: Jean-Denis Durou (IRIT, Université de Toulouse) , Maurizio Falcone (Dipartimento di Matematica, Università di Roma "La Sapienza") .

Authors:

Björn Häfner (Technical University of Munich)

Keywords:

bayesian methods, computer vision, image enhancement, inverse problems, nonlinear optimization