Trimble Harrier Airborne laser scanning point cloud over the town of Biberach, Germany displayed in Inpho DTMaster, courtesy of Trimble.
Trimble Harrier Airborne laser scanning data over the town of Biberach, Germany displayed as TIN surface in Inpho DTMaster, courtesy of Trimble.
3D view of electrical substation outside of Denver, Colorado (colored by elevation), courtesy of Merrick.
2D view of electrical substation outside of Denver, Colorado (colored by elevation), courtesy of Merrick.
An image generated in GeoDigital’s VegWorks mobile work management software reflects asset and all related field data including poles, spans, maps, aerial photography, LiDAR mapping, customer information and work history. The application merges the data to allow a utility to quickly identify any right-of-way threats and to produce all work order details to dispatch crews for remediation.
GeoDigital’s stabilized airborne and terrestrial remote sensing systems are able to acquire base map data to produce three-dimensional images of right-of-way and structures. They combine traditional corridor imagery and asset mapping technologies with a cost-effective LiDAR to survey the transmission line for clearance problems, danger vegetation and abnormal geometry, while providing access to engineering survey grade LiDAR when needed.
The market for software to manage, process, fuse, and visualize LiDAR point cloud data is still very young and amorphous. As both the amounts of LiDAR data and the demand for it continue to grow, software vendors are developing innovative solutions to enable their clients to integrate this data into their workflows.
Large software developers (such as ESRI, Intergraph, and Autodesk) now support LiDAR natively in their software. Some manufacturers of laser scanners (such as Riegl, Leica Geosystems, and Optech) are expanding the functionalities of the software applications that they provide with their hardware, so as to try to keep their customers in their fold. Another manufacturer, Trimble, makes its integrated terrest-rial solution available only in combination with its hardware, while it sells its airborne software to users without restrictions, because it wants to maintain the user base to which it has freely sold its eCognition and Inpho products. Some large users of LiDAR data such as GeoDigital and Merrick have developed their own in-house software, which they either keep proprietary (as in the case of GeoDigital) or sell (as in the case of Merrick’s MARS software suite). Free, open-source LiDAR software, such as PDAL is also available.
Editor’s Note: This is a comprehensive update on lidar software, with interviews of several key companies. Click here for more information about the offering from Exelis (formerly ITT VIS). Click here for more information about LiDAR QA/QC.
LiDAR continues to grow in popularity, especially throughout the architecture, engineering, and construction world. The fact that all the major software vendors are now supporting it natively is clear evidence that it has become widely accepted. This is opening a much larger market for LiDAR data and how it is being used, says Bill Emison, Senior Account Manager at Merrick. “Unfortunately, however, while there is great variety available, there is no consistency.” In fact, the market for LiDAR software “is still an open game,” says Michael P. Gerlek, an independent consultant with Flaxen Consulting. “There are many different players and they are trying many different things. LiDAR software now is at roughly the same development stage as software was for raster data 10-15 years ago.”
“Hardware manufacturers,” Gerlek explains, “are focusing on providing the specific software tools that they need for the data coming off of their devices, such as calibrating and corrections. Some of them are trying to provide additional pieces. Independent, third-party developers are going to develop more robust tools that work with post-processed data from different sensors. End users do the best that they can with the available tools, plus they develop their own.”
In aerial LiDAR, which has been around much longer than ground-based LiDAR, users typically piece together applications and there are already many systems on the market. “Consolidation might happen for specific applications, such as corridor mapping, but there will always be a place for more open systems,” says Gregor Willhauck, Software Product Line Manager for Trimble’s GeoSpatial Division, where he manages a team of product managers for eCognition, Inpho and Trident Analyst. By contrast, he points out, in terrestrial LiDAR, there is more of a trend toward integrated solutions.
Some firms have increased their capabilities in this area through acquisitions. In 2010, Hexagon bought first ERDAS, which had developed its own LiDAR software tools, and then Intergraph. In October 2011, Autodesk bought Netherlands-based Alice Labs, which specializes in editing and displaying point cloud data. At the beginning of December 2011, GeoDigital, which specializes in the use of LiDAR for vegetation management and line rating and clearance analysis, acquired Powel, Inc., maker of a mobile work management product line that includes an application for managing utility field design, vegetation maintenance, outage recovery, and mapping. A notable exception to this trend is GIS giant Esri, which did not acquire a company specifically for LiDAR.
Not everybody, however, thinks that consolidation is the best way forward. “For managing, processing, and visualizing LiDAR data, I’d rather have separate packages, because that provides more flexibility,” says Gerlek. “Software packages should focus on what they are good for, then stop.” He points out that the available tools are, by and large, focused on specific types of analysis, such as feature extraction or linear and volumetric measurement, but there are few general software development kits (SDKs).
The biggest trend you will see is a move to the creation of ultra-high density data and to processing on demand. For example, you can scan a complete oil refinery...but a client who just wants to replace two pipes does not need that... So, the operating principle should be scan everything, process as needed.
– Alastair Jenkins of GeoDigital
Formats and Standards
For the whole industry to grow, hardware manufacturers, software developers, and end users need interoperable formats. In the terrestrial market, the standards typically come from surveying and are more proprietary than in the aerial market, says Willhauck. The closest thing the industry has to a standard format is the LAS, a specification published, maintained, and copyrighted by the American Society for Photogrammetry and Remote Sensing (ASPRS). Developed primarily for exchange of LIDAR point cloud data, the LAS is a public, binary file format for the interchange of any 3D point cloud data. However, according to Alastair Jenkins, President and CEO of GeoDigital, the LAS “has led to a massive war between companies to suit their own prejudices and advantages. The latest version still has a way to go.”
“Companies that are end users of LiDAR data don’t want to have to write their own file format support; they want that for free,” argues Gerlek. “In general, having many different file formats out there incurs a development and maintenance burden on the vendor. Users need to be able to collect data with a sensor from one manufacturer, process it with software tools from a second manufacturer, and visualize it with software from a third one. Standards smooth out those transitions and allow different vendors to develop their own stuff and play together nicely. The days of closed, proprietary software are dead. The world no longer needs it. My former employer, LizardTech, is getting very little traction in the LiDAR space, in large part because the value of openness outweighs other possible feature advantages.”
As reported in previous articles on LiDAR in this magazine, the growth in data from LiDAR sensors continues to outpace the ability of users’ software to manage it and process it. “That’s not going to change any time soon,” says Gerlek. Yet the ability to manage very large files is critical for production shops. To deal with this challenge, heavy users of LiDAR data leverage GPU processing, compress and index data, and use 64-bit operating systems, which allow them to use more RAM and multiple CPUs. In particular, “quite a bit of research” is going into data indexing, says Emison.
Jenkins points out a different challenge: figuring out an efficient way to quality-control the data. “Five years ago,” he says, “you could process 70 square kilometers in four hours and then you would spend 10 to 15 minutes doing QA/QC. Now you can process that data automatically in 10 to 15 minutes, but QA/QC will take you four hours. The real problem today is this: either you accept the errors or, if you need a certain degree of accuracy, you need to figure out an efficient way to do the QA/QC.”
Gerlek predicts that ArcGIS, SOCET SET, MicroStation, etc., eventually will incorporate point clouds to the same degree that they currently support raster data. While some of these software products already have 3D support today, he points out, they are typically limited in functionality and cannot support very large models and analysis.
Will LiDAR follow the current trend toward cloud-based applications and data storage? “Since the majority of LiDAR data processing utilizes LAS (flat) files,” says Emison, “I believe that the trend will support a migration from LAS files to relational databases (such as Oracle and SQL Server), then to cloud-based applications. With the pulse rate growing and the point densities rising, most organizations will have to develop a data storage system to handle massive amounts of data. Furthermore, as LiDAR becomes more popular, demand for this type of data will increase.”
Trimble does not have any cloud/Web-based solutions, says Willhauck, but he sees “a trend toward it.” He is not at liberty to comment on whether his company is currently developing any such solutions, but he points out that Trimble has other cloud-based businesses.
Going forward, most LiDAR software will continue to be specialized, Emison opines. “I don’t see a single package but more software growing out of the verticals, not just a one-size-fits-all solution. I’d like to think ours would be that, but I don’t think it will be the only game in town, the way Esri is for GIS.”
A close look at Google Maps reveals many lines that do not connect properly. Whether that matters to users depends on their reason for using the system. For personal navigation, for example, it does not matter, but for engineering purposes it does. “Now there is a very broad spectrum of needs with regards to geographic accuracy,” Jenkins points out. “The biggest trend you will see is a move to the creation of ultra-high density data and to processing on demand. For example, you can scan a complete oil refinery and use that data to make a completely new as-built, but a client who just wants to replace two pipes does not need that. If you have the model, you can make it easy for your clients to make changes. So, the operating principle should be scan everything, process as needed.”
Much very promising research is being done with regard to fusing raster/vector data with 3D, but 3D will still be processed independently for the next couple of years, says Gerlek.
Trimble has a complete land-mobile LiDAR solution, which is a combination of software and hardware, and an aerial LiDAR system that matches with its Inpho suite. “Because we have both air and land within the same division, we are working to apply the same solutions to both, to navigate and store point clouds,” says Willhauck. “These solutions also extend into Trimble’s other divisions.”
Trimble has a strong focus on automating the process of feature extraction. While the Inpho suite has a strong offering of LiDAR filtering algorithms (developed in cooperation with the University of Vienna), eCognition provides state-of-the-art object-based feature extraction capabilities. Between the two software packages, the complete LiDAR workflow from data processing and filtering to feature extraction is covered. All of Trimble’s LiDAR systems come with an optical component: an on-board camera that offers a quick solution to colorizing point clouds. This combination benefits automatic feature extraction and enables eCognition to combine LiDAR data and raster imagery into a GIS. See Figures 1-2.
Merrick is an engineering firm and its geospatial group is responsible for collecting LiDAR data, processing it, and producing derivative products. “Since we’ve been in the collection game for about a decade, we decided to build our own system,” says Emison. “That is why we developed MARS and use it through our entire workflow process. We do use other software, most notably ESRI and some imagery products for ortho photos. We also offer MARS commercially – mostly for those who need basic tools to manage and visualize data, chop it up, re-project it, etc. – and offer training. Our goal is not to be a major software vendor but to use it, enhance it as needed, and sell it as we can.”
According to Emison, MARS is focused mostly on production, while Esri’s software is focused on end users. “We build software for people who need to manage LiDAR data, while they build software for those who just need to use it and analyze it. We are building vertical applications that support different industries, such as mapping power lines for vegetation clearance safety analysis or mapping obstructions of in/out-bound airport paths, as well as more specialized packages for forestry, hydromodeling, etc.” See Figures 3-4.
GeoDigital is a LiDAR acquisition company that specializes in corridor mapping, using engineering-grade systems mounted on helicopters, fixed-wing aircraft, and land vehicles. It produces 3D models, with an absolute vertical accuracy of less than 10 centimeters and a horizontal accuracy of less than 15 centimeters, which organizations can use to map, assess, and understand their assets. “We are scanning 7,000 miles of power lines per month, which used to be a year’s capacity for a large firm,” says Jenkins. “Our internal software group doubled in size when we acquired Powell.”
“We use data from an inertial measurement unit (IMU) and a GPS receiver to determine the collection platform’s position in space and its orientation; we then combine that data with the range to the target to determine the polar coordinates of each return. This is the hardest step to do very accurately. If you don’t start off with good data, it never gets better.” There are commercial off-the-shelf solutions to process the inertial and GPS data, such as Applanix’ POS product line. “We have done very significant writing of our own software, to achieve higher accuracy, make a very accurate point cloud, and fuse it with imagery. We don’t sell it because it is part of our core offering.”
PDAL – The Evolution of an Open Source Option
Various tools have been developed to manage and manipulate LAS files. Howard Butler developed libLAS, a C/C++ library, for the Iowa Department of Natural Resources for use in its statewide LiDAR project and for general use. While libLAS is used for production purposes in the real world, explains independent consultant Michael Gerlek of Flaxen Consulting, it is limited to the LAS file format. He and Butler wanted a tool that could do more, such as provide a low-level file interchange capability and basic operations. So they developed PDAL as a successor to libLAS and as the 3D version of GDAL, another format with which they have also been involved and that has been around for years. Like libLAS, PDAL has received funding from the U.S. Army Corps of Engineers, which had additional requirements that called for a whole new library. “PDAL fills a market need that nobody else is filling, has no competitors, and is not competing against any product or company,” says Gerlek. “We provide an underlying library on top of which others can build, if they need to do transformations or filtering pipelines.”
GeoDigital’s primary visualization tool is its Grid Intel Online (GIO) Web application, which allows its clients to access billions of points on demand, via a product like Google Earth. They can draw a box around their area of interest and download the point cloud for that area. “Our strength is that we provide our software in the form of a viewing tool, which is customized for handling massive amounts of data and allows such functions as zooming,” says Jenkins. “Our focus is to make that data available faster and faster.”
When it bought Powell earlier this Winter, GeoDigital turned it into a new software and analytics division, GeoDigital Solutions, Inc., with responsibility for software development and the integration of complementary technologies into turnkey enterprise solutions for utilities and other asset-intensive indust-ries, including rail, oil and gas, transportation, and telecommunications.
“We transform LiDAR data into something that is useful for a utility,” says Scott Rogers, the division’s President. “We have been building tools for the utility industry for about 15 years. We provide field tools to identify features, document them with a quick user interface, and report back to the rest of the organization. This is very important for meeting regulatory requirements. We turn LiDAR data into actionable information and assure that the work has been completed. Our software can generate a cost estimate and feed back-end systems. We are selling both the software and the service, depending on our clients’ needs.” See Figures 5-6.
Esri has been working with LiDAR data for many years and has partners that process it. In the past, LiDAR data has been used mainly for elevation, explains Steve Snow, Esri Marketing Specialist. Users imported LiDAR data into a geodatabase and then built terrain datasets. “With ArcGIS 10.1, users will be able to directly read LAS files through the new LAS Dataset. They will be able to render LiDAR data on the fly as a surface or as a point cloud.”
Esri’s 3D team developed these capabilities in house. This year, the company will provide a series of free LiDAR workshops around the country. “We are doing this to support decision makers,” says Snow. “They want faster access to the information via GIS, so that they can make timely and accurate decisions. A point cloud has no value until you extract the information from it. All geospatial software is moving to 3D. It is now an expectation.”
Point clouds are no longer restricted to LiDAR. Around the middle of 2012, Microsoft’s Vexcel Imaging division, which manufactures digital aerial cameras, plans to release software that generates point clouds by using overlapping aerial images. “We are still in the development phase and only run the software on development machines,” says Alexander Wiechert, the division’s business director. “We fly an area with our UltraCam, firing it extremely fast, and get highly overlapped aerial images. Then we do dense matching of each pixel. We achieve much higher point densities than with LiDAR. It is a new trend, but we have a lot of relevant experience modeling cities.”
Software vendors, end users, and academics are developing new tools to manage and process LiDAR point cloud data. While it is too early to know whether any one particular software packet will dominate the market in a few years, it is clear that LiDAR data will soon be as standard and indispensable a tool for geospatial analysis as raster data is today. Additionally, the availability of vast amounts of point cloud data is greatly boosting the development of 3D modeling software.