Daisy filter flow: A generalized discrete approach to dense correspondences

Establishing dense correspondences reliably between a pair of images is an important vision task with many ap-plications. Though significant advance has been made to-wards estimating dense stereo and optical flow fields for two images adjacent in viewpoint or in time, building re-liable dense correspondence fields for two general images still remains largely unsolved. For instance, two given im-ages sharing some content exhibit dramatic photometric and geometric variations, or they depict different 3D scenes of similar scene characteristics. Fundamental challenges to such an image or scene alignment task are often mul-tifold, which render many existing techniques fall short of producing dense correspondences robustly and efficiently. This paper presents a novel approach called DAISY filter flow (DFF) to address this challenging task. Inspired by the recent PatchMatch Filter technique, we leverage and extend a few established methods: 1) DAISY descriptors, 2) filter-based efficient flow inference, and 3) the Patch-Match fast search. Coupling and optimizing these mod-ules seamlessly with image segments as the bridge, the pro-posed DFF approach enables efficiently performing dense descriptor-based correspondence field estimation in a gen-eralized high-dimensional label space, which is augmented by scales and rotations. Experiments on a variety of chal-lenging scenes show that our DFF approach estimates spa-tially coherent yet discontinuity-preserving image align-ment results both robustly and efficiently. 1