This paper describes Google Street View, a global-scale project capturing and serving panoramic street-level imagery using custom vehicles and camera systems, discussing technical challenges in data collection, processing, pose optimization, and 3D data leveraging for smart navigation.
Main Contributions
Details the evolution of Google Street View vehicular platforms from initial prototypes to current R5 and R7 camera systems.
Explains the iterative engineering approach used in developing and operating data collection vehicles globally.
Describes the pose optimization process, combining GPS, wheel encoder, and inertial data with map snapping for accurate navigation.
Introduces smart navigation features leveraging 3D data, including click-to-go and depth map creation for immersive user experiences.
Outlines methods for serving imagery to users, including panorama stitching, tiling, and privacy protection through face and license plate detection.
Abstract
Street View serves millions of Google users daily with panoramic imagery cap-tured in hundreds of cities in 20 countries across four continents. A team of Google researchers describes the technical chal-lenges involved in capturing, processing, and serving street-level imagery on a glob-al scale.