Location-aware applications play an increasingly critical role in everydaylife. However, the most common global localization technology - GPS - haslimited accuracy and can be unusable in dense urban areas and indoors. Weintroduce an image-based global localization system that is accurate to a fewmillimeters and performs reliable localization both indoors and outside. Thekey idea is to capture and index distinctive local features in ground textures.This is based on the observation that ground textures including wood, carpet,tile, concrete, and asphalt may look random and homogeneous, but all containcracks, scratches, or unique arrangements of carpet fibers. These imperfectionsare persistent, and can serve as local features. Our system incorporates adownward-facing camera to capture the fine texture of the ground, together withan image processing pipeline that locates the captured texture patch in acompact database constructed offline. We demonstrate the capability of oursystem to robustly, accurately, and quickly locate test images on various typesof outdoor and indoor ground surfaces.
展开▼
