Wide-baseline stereo from multiple views: A probabilistic account

This paper describes a method for dense depth reconstruction from a small set of wide-baseline images. In a wide-baseline setting an inherent difficulty which complicates the stereo-correspondence problem is self-occlusion. Also, we have to consider the possibility that image pixels in different images, which are projections of the same point in the scene, will have different color values due to non-Lambertian effects or discretization errors. We propose a Bayesian approach to tackle these problems. In this framework, the images are regarded as noisy measurements of an underlying 'true' image-function. Also, the image data is considered incomplete, in the sense that we do not know which pixels from a particular image are occluded in the other images. We describe an EM-algorithm, which iterates between estimating values for all hidden quantities, and optimizing the current depth estimates. The algorithm has few free parameters, displays a stable convergence behavior and generates accurate depth estimates. The approach is illustrated with several challenging real-world examples. We also show how the algorithm can generate realistic view interpolations and how it merges the information of all images into a new, synthetic view.