Reconstruction of 3D Models from Intensity Images and Partial Depth

Abstract

This paper addresses the probabilistic inference of geometric structures from images. Specifically, of syn-thesizing range data to enhance the reconstruction of a 3D model of an indoor environment by using video images and (very) partial depth information. In our method, we interpolate the available range data us-ing statistical inferences learned from the concurrently available video images and from those (sparse) regions where both range and intensity information is avail-able. The spatial relationships between the variations in intensity and range can be efficiently captured by the neighborhood system of a Markov Random Field (MRF). In contrast to classical approaches to depth re-covery (i.e. stereo, shape from shading), we can af-ford to make only weak prior assumptions regarding specific surface geometries or surface reflectance func-tions since we compute the relationship between exist-ing range data and the images we start with. Experimen-tal results show the feasibility of our method.