Recent advances in geospatial data acquisition technologies are enabling automatic urban reconstruction in ever increasing scale. Though a number of mature systems are developed for interative exploration of the final reconstructed urban environments, few try to visualize the intermediate products of the entire reconstruction pipeline. We present a system to interactive navigate and explore large scale reconstruction data sets consisting of input images, 3D point clouds and polygonal meshes in an integrated Internet-based platform. The massive scale of data is handled by hierachical level of details methods, which generates simplified data through mesh simplification and point resampling as well as streams and visualizes data progressively in a fine-grained manner. We employ cac...[ Read more ]

Recent advances in geospatial data acquisition technologies are enabling automatic urban reconstruction in ever increasing scale. Though a number of mature systems are developed for interative exploration of the final reconstructed urban environments, few try to visualize the intermediate products of the entire reconstruction pipeline. We present a system to interactive navigate and explore large scale reconstruction data sets consisting of input images, 3D point clouds and polygonal meshes in an integrated Internet-based platform. The massive scale of data is handled by hierachical level of details methods, which generates simplified data through mesh simplification and point resampling as well as streams and visualizes data progressively in a fine-grained manner. We employ cache, culling and compression algorithms specifically tailored to our hierachical data structure to further speed up rendering. Our system is built directly into standard browser, which can be accessed independent of platform including mobile devices with Internet access. Our system has hosted a dozen of large scale urban reconstruction datasets and we will demonstrate several representative datasets in this thesis.