Content pipeline from LIDAR to Unreal engine (2015)
In this work a set of LIDAR scan sites from Riegl are converted from point cloud format into RGB-D images, whose geometry is dropped from high-resolution triangulation into shape preserving low-poly format. After that rough stitching is performed in 3D and resulting textured mesh pieces are combined together and loaded into Unreal 3D engine for interactive visualization.
3D point cloud data along with image and depth images illustrated.
Rough 3D stitching perfomed by selecting largest and most visible geometry vs scan sites.
Automatic camera settings cause large variations in surface coloring and brightness across the scan sites.
Depth map simplification starts from Delaunay triangulated point projection.
Vertices are removed iteratively by minimizing the geometrical loss.
Finally the high-resolution mesh is represented in low poly format.
Low poly model loaded into Unreal 3D.
Low poly model loaded into Unreal 3D.
Low poly model loaded into Unreal 3D.
Low poly model loaded into Unreal 3D.
Low poly model loaded into Unreal 3D.
Low poly model loaded into Unreal 3D.
Low poly model loaded into Unreal 3D.
