CAREER: Photorealistic 3-D Modeling of Large-Scale Scenes: Integration of 3-D Range and 2-D Intensity Sensing in a Complete System

该项目的目标是自动重建大规模 3D 场景的高度准确的真实感描述，同时最大限度地减少人类交互并提高计算效率。激光测距传感技术的最新发展使得获取大范围场景的精确 3D 扫描成为可能。将利用密集且复杂的 3D 范围数据集以及高分辨率彩色图像（地面和空中）。研究将在范围分割（为了处理几何复杂性）、范围和 2D 图像之间的自动配准、3D 建模、航空摄影融合和可视化等领域进行。 拟议的研究将直接影响城市规划、城市设计、防灾、电子新闻、建筑、考古、历史保护和娱乐等领域的应用。 将开发的新颖解决方案将对 PI 的教育议程及其外展活动产生直接影响。 计算机科学、地理、电影和媒体系之间的交叉融合是另一个优先事项。最后，与博物馆等机构以及专注于 3D 场地建模和场地建设的公司的合作是将这项研究走出学术环境的重要一步。

The goal of this project is the automated reconstruction of highly accurate photorealistic descriptions of large-scale 3D scenes with a minimum of human interaction and increased computational efficiency. Recent developments in laser-range sensing technology make possible the acquisition of accurate 3D scans of large-scale scenes. Dense and complex 3D range data-sets along with high resolution color imagery (ground and aerial-based) will be utilized. Research will be carried in the areas of range segmentation (in order to handle the geometric complexity), automatic registration between range and 2D images, 3D modeling, fusion of aerial photography, and visualization. The proposed research is going to directly affect applications in the areas of urban planning, urban design, disaster prevention, electronic journalism, architecture, archeology, historical preservation, and entertainment. The novel solutions that will be developed will have a direct effect on the educational agenda of the PI and on his outreach activities. The cross-fertilization between Computer Science, Geography, and Film and Media departments is another priority. Finally, collaboration with institutions such as museums, and with companies concentrated on 3D site modeling and site construction is an essential step for forwarding this research out of the academic environment.