LiDAR data is a significant resource for identifying similar geospatial features in urban planning, land use analysis, emergency response, and other applications. Traditionally LIDAR is analyzed through manual process, which is a very challenging task due to the need to identify similarities over a growing size and complexity of data. To alleviate this challenge, we designed and developed a GPU-powered visual image analytics system to handle this operation at large scale. Our system encodes human-freeform-LiDAR selection into 2D images through an autonomous image analysis process that matches selected areas of interest. To ensure the system's practicality in handling hundreds of stitched LiDAR patches, we have scaled up our algorithms through a series of parallelized GPU processing, analyzing, and encoding methods. We conducted informal user studies to assess the utility and usability of the system.
Todd Eaglin, Xiaoyu Wang, Bill Ribarsky, "Interactive Visual Analytics in Support of Image-Encoded LiDAR Analysis" in Proc. IS&T Int’l. Symp. on Electronic Imaging: Visualization and Data Analysis, 2016, https://doi.org/10.2352/ISSN.2470-1173.2016.1.VDA-495