Belief propagation optical flow for high-resolution image morphing

Over the last decade, considerable progress has been made on the so-called early vision problems. We present an optical flow algorithm for image morphing that incorporates recent advances in feature matching, energy minimization, stereo vision and image segmentation.

At the core of our flow estimation we use Efficient Belief Propagation for energy minimization. While state-of-the-art algorithms only work on thumbnail-sized images, our novel feature downsampling scheme in combination with a simple, yet efficient data term compression can cope with high-resolution data. The incorporation of SIFT features into data term computation further resolves matching ambiguities, making long-range flows possible. We detect occluded areas by evaluating the symmetry of the flow fields, we further apply Geodesic matting to automatically inpaint these regions.