mdBrief - A Fast Online Adaptable, Distorted Binary Descriptor for Real-Time Applications Using Calibrated Wide-Angle Or Fisheye Cameras

摘要

Fast binary descriptors build the core for many vision based applicationswith real-time demands like object detection, Visual Odometry or SLAM. Commonlyit is assumed, that the acquired images and thus the patches extracted aroundkeypoints originate from a perspective projection ignoring image distortion orcompletely different types of projections such as omnidirectional or fisheye.Usually the deviations from a perfect perspective projection are corrected byundistortion. Latter, however, introduces severe artifacts if the camerasfield-of-view gets larger. In this paper, we propose a distorted and maskedversion of the BRIEF descriptor for calibrated cameras. Instead of correctingthe distortion holistically, we distort the binary tests and thus adapt thedescriptor to different image regions.