The Photogrammetry, 3D Visualization, and Lidar (P3DL) Conference aims to drive innovation and explore capabilities for precision geopositioning/targeting, precision mensuration, 3D visualization, lidar, and 3D data-related technologies to address the needs of the U.S. Department of Defense and Intelligence Community. Attendees include professionals from U.S. government, academia, and industry.
Kitware is looking forward to presenting at P3DL. Titled, ‘Open Source Software for Deriving Meshes from Point Clouds with Error Propagation,’ this presentation will provide an update on Kitware’s progress toward building free and open source software for deriving geospatial mesh models from imagery and point cloud sources. Please see below for more information.
Open Source Software for Deriving Meshes from Point Clouds with Error Propagation
Date and time TBD
Presented by Matt Leotta, Ph.D., a technical leader on Kitware’s Computer Vision Team
In this talk, we will give an update on Kitware’s progress toward building free and open source software for deriving geospatial mesh models from imagery and point cloud sources. Our prior work on the IARPA CORE3D program led to the release of Danesfield, an open source software system that builds on Raytheon’s P3D software to reconstruct 3D textured mesh models of buildings and terrain. In recent months, we have extended this platform in several ways. First, we have integrated VisSat as an alternative to P3D. VisSat is an open source P3D replacement developed by a competing team on CORE3D. Thus, we now have a completely open CORE3D pipeline that can be used freely without licensing restrictions. Second, we have generalized our Danesfield pipeline to process point clouds from other sources, such as lidar or FMV. Our open source TeleSculptor software provides FMV to point cloud conversion. Third, we have developed open source tools to convert the Danesfield mesh outputs to 3D Tiles for efficient web-based mesh visualization. Finally, we have started preliminary work to incorporate error model propagation in the point cloud to mesh conversion process. We are working to extend the error propagation standards set forth in the NGA Generic Point-cloud Model (GPM) to meshes, and to propagate errors from P3D point clouds or other sources into the final mesh representation. We will describe our approach to propagating error (covariance) to meshes and show the results of preliminary work.