Integration of Mobile and Aerial LiDAR for Large-Scale Photorealistic Urban Modeling

点击翻译按钮获取中文摘要

The interest in generating 3D models has been motivated by a wide range of applications such as urban planning and GIS. In recent years, the demand for 3D photorealistic building models has dramatically increased. These needs arise particularly from 3D GPS navigation systems and on-line services such as Nokia, Google, and Apple maps. According to Parks Associates (2010), the number of Smartphone users worldwide is predicted to exceed 1 billion by 2014. The mobile location based services (MLBS) including advertising and marketing is to generate multi-billion revenues, as a recent report from Juniper Research showing that the market of MLBS is expected to reach more than $12.7 billion by 2014. Maps are the backbone of the location based services, where the three dimensional maps are one of the most important features in the next generation maps. To date, no efficient solutions are able to fully automate the creation of photorealistic building models with ground-level details from LiDAR (Light Detection And Ranging). Automatic methods could significantly increase productivity, reduce costs, and be of enormous interest to location based service providers. Mobile LiDAR is a rapidly emerging technology in which multiple sensors mounted on a vehicle collect 3D and visual data while vehicles drive at the posted speed. The ability to produce accurate geo-referenced 3D points with sufficient detail at ground level makes mobile LiDAR an ideal data source for detailed façade modeling. Yet, due to the line-of-sight issue of ground-based data acquisition system, it is difficult to always obtain complete and sufficient data from all the building surfaces such as rooftops. The combination of mobile and aerial LiDAR will contain information from both a top-down perspective and a street-level view, which therefore is a valuable data source for large-scale photorealistic urban modeling. This proposal investigates three key issues in automatic creation of 3D photorealistic building models at city-scale using mobile and aerial LiDAR: 1) automatic mobile-to-aerial LiDAR registration, 2) automatic photorealistic building model generation, and 3) automatic texture selection and mapping.