Abstract

We propose a novel approach that generates a high-quality depth map from a set of images captured with asmall viewpoint variation, namely small motion clip. Asopposed to prior methods that recover scene geometry andcamera motions using pre-calibrated cameras, we introduce a self-calibrating bundle adjustment tailored for small motion. This allows our dense stereo algorithm to producea high-quality depth map for the user without the need forcamera calibration. In the dense matching, the distributionsof intensity profiles are analyzed to leverage the benefit ofhaving negligible intensity changes within the scene due tothe minuscule variation in viewpoint. The depth maps ob-tained by the proposed framework show accurate and ex-tremely fine structures that are unmatched by previous literature under the same small motion configuration.