A stochastic optimization approach to stereo matching is presented. Unlike conventional correlation matching and feature matching, the approach provides a dense array of disparities, eliminating the need for interpolation. First, the stereo matching problem is defined in terms of finding a disparity map that satisfies two competing constraints: (1) matched points should have similar image intensity, and (2) the disparity map should be smooth. These constraints are expressed in an '(energy" function that can be evaluated locally. A simulated annealing algorithm is used to find a disparity map that has very low energy (i.e., in which both constraints have simultaneously been approximately satisfied). Annealing allows the large-scale structure of the disparity map to emerge at higher temperatures, and avoids the problem of converging too quickly on a local minimum. Results are shown for a sparse random-dot stereogram, a vertical aerial stereogram (shown in comparison to ground truth), and an oblique ground-level scene with occlusion boundaries.
This paper describes a highly successful application of MRFs to the problem ofgenerating high-resolution range images. A new generation of range sensors combines the capture of low-resolution range images with the acquisition of registered high-resolution camera images. The MRF in this paper exploits the fact that discontinuities in range and coloring tend to co-align. This enables it to generate high-resolution, low-noise range images by integrating regular camera images into the range data. We show that by using such an MRF, we can substantially improve over existing range imaging technology.
This paper addresses the probabilistic inference of geometric structures from images. Specifically, of synthesizing 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 using statistical inferences learned from the concurrently available video images and from those (sparse) regions where both range and intensity information is available. The spatial relationships between the variations in intensity and range can be efficiently captured by the neighborhood system of a Markov Random Field (MRF).
Lowry, L. Quam, G. Smith, and A. Witkin SRI International, Menlo Park, California 94025 ABSTRACT This paper describes the results obtained in a research program ultimately concerned with deriving a physical sketch of a scene from one or more images. Our approach involves modeling physically meaningful information that can be used to constrain the interpretation process, as well as modeling the actual scene content. In particular, we address the problems of modeling the imaging process (camera and illumination), the scene geometry (edge classification and surface reconstruction), and elements of scene content (material composition and skyline delineation). I INTRODUCTION Images are inherently ambiguous representations of the scenes they depict: images are 2-D views of 3-D space, they are single slices in time of ongoing physical and semantic processes, and the light waves from which the images are constructed convey limited information about the surfaces from which these waves are reflected. Therefore, interpretation cannot be strictly based on information contained in the image; it must involve, additionally, some combination of a priori models, constraints, and assumptions.