Integrating LIDAR Range Scans and Photographs with Temporal Changes

Registering 2D and 3D data is a rapidly growing research area. Motivating much of this work is the fact that 3D range scans and 2D imagery provide different, but complementing information about the same subject. Combining these two perspectives leads to the creation of accurate 3D models that are texture mapped with high resolution color information. Imagery can even be obtained on different days and in different seasons and registered together to show how a scene has changed with time. Finding correspondences among data captured with different cameras and containing content and temporal changes can be a challenging task. We address these difficulties by presenting a contextual approach for finding 2D matches, performing 2D-3D fusion by solving the projection matrix of a camera directly from its relationship to highly accurate range scan points, and minimizing an energy function based on gradient information in a 3D depth image.