Issue link: https://read.dmtmag.com/i/103965
loss: LASzip from rapidlasso and Lidar Compressor from LizardTech. LASzip is based on predictive techniques and entropy encoding, and it claims compression rates in the 4x to 13x range (over uncompressed LAS files), with compression and decompression rates in the millions of points per second. The open-source aspect of LASzip is especially valuable in light of emerging opendata policies. The National Land Survey of Finland, for example, has published its entire open LiDAR inventory in the LASzip (LAZ) format. LASzip also won the 2012 Geospatial World Forum Technology Innovation award in LiDAR Processing. LiDAR Compressor is an extension of LizardTech's MrSID raster image-compression technology, which accommodates point-cloud data and makes LiDAR data directly exploitable in a compressed form. MrSID is a wavelet-based compression technology that leverages the redundancies occurring over resolutions levels. This focus on resolution means that the aforementioned pre-pyramiding is an inherent part of the file format, enabling very fast low-resolution decodes without redundant storage or "ingestion" time. Similarly, the format incorporates a spatial index to enable direct access to points in the extent of interest. These compression solutions are being embraced by the industry with several mainstream applications adopting one or both (e.g. ArcGIS, Merrick MARS, Global Mapper, Safe Software's FME, Overwatch's Lidar Analyst and several others). Server Solutions Another approach to the LiDAR size issue is making just the region of interest available via client/server interaction. The good news is that this is potentially very fast. The computational overhead associated with decoding compressed data can be significant, and it's nonexistent if there's no compression. The bad news is that it requires a lot of hardware and networking infrastructure. Fortunately, such solutions now are increasingly feasible. This is exactly the tradeoff that Google Earth and Bing Maps made with raster data: there's little sophisticated compression going on there—indeed there's tremendous duplication and redundancy—but they've overwhelmed the problem with hardware, and the result is a great user experience. Though clearly in its infancy, LiDAR server technology seems to be emerging. QCoherent (now owned by GeoCue), Dielmo and Esri each make Web servers that expose LiDAR data. The servers host the raw point-cloud data, expose various visualizations (e.g., contours, elevation-colored points, profiles, etc.) and enable exporting of the raw data associated with a given extent. Imagery/LIDAR Special Issue Figure 3. A colorized point-cloud layer is draped over a shaded-relief base map served by QCoherent's LiDAR Server. Open Source There's one further thing to keep an eye on: opensource innovations. Open-source libraries and applications such as GRASS, QGIS and GDAL form a large portion of the industry's foundation. They don't do much with LiDAR, but that's starting to change. A few years ago, LizardTech contributed a driver to the GDAL project that creates DEMs for LiDAR files compressed to its MG4 format. Shortly after that, libLAS, an open-source project with an initial focus of reading the LAS data format, broadened its horizons to include multiple formats and basic 3-D point-cloud manipulation. Since renamed PDAL (Pointcloud Data Abstraction Layer), it's now supported by some of the same core developers that maintained GDAL. Meanwhile, Martin Isenburg, the author of LASzip, maintains a suite of open-source tools for LiDAR processing called LASTools that continues to be quite popular. As raster data continue to be gathered in imagery of higher resolution, LiDAR data mount in point clouds of increasing density and extent. The evolution of sensor technologies through recent decades has left plenty of opportunity in its wake for innovation in ways to exploit the captured data. Getting those innovations into mainstream geospatial applications will be the ongoing work for many years to come. Mike Rosen is software architect and Matt Fleagle is senior technical writer, LizardTech; e-mail: mrosen@lizardtech.com and mfleagle@lizardtech.com, respectively. J A N U A R Y 2 O 1 3 / W W W . G E O P L A C E . C O M 21