A 2.942Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
LiDAR technologies are becoming more and more usable in diverse areas of human activity. The application of LiDAR data varies from heavy industry to environmental protection, infrastructure to navigation and architecture to agriculture.
So LiDAR data appears widely populated, but the current users often experience unexpected difficulties, disappointment and frustration as they are not prepared for the fact that LiDAR data not only brings opportunities but challenges too. All data must be used in a proper way and interpreted with correct means in order to obtain anticipated results with the desired level of accuracy.
The natural beauty of high-resolution LiDAR point clouds and the impression of virtual reality which they create make them inexplicably and magically attractive for people who very often can’t clearly realize how to use the collected LiDAR data in an effective way. So one may often notice the phenomenon where the engineering departments buy LiDAR equipment and/or LiDAR services in order to inspect their facilities (plants, railways, buildings, roads, open pit mines, underground mining objects etc.) without a clear understanding that the obtained point cloud, "as is", often does not give resolution to their actual problems.
This especially relates to both mobile and static 3D LiDAR since airborne LiDAR data has already become a commodity with certain standardized processing work flows, however it is worth mentioning usually it is still not an optimal and effective solution.
.Typically laser equipment producers (FARO, RIEGL, Leica, Z+F, Neptec, Optec etc.) sell the scanners with some accompanied in-house software which is usually limited to viewer, a basic georeferencing tool and export capabilities from proprietary scanner format to some interchange LiDAR format (i.e. .las or .pts). These generic features in most cases are not able to provide the necessary data processing services for any customer with non-standard specialized needs. In other words buying a good scanner, making good scans of your problematic areas (whatever they are) and then analysing these scans with the help of a generic laser-supplier’s own software does not take the end-user any closer to the resolution of his current problems.
Having been actively involved in LiDAR custom data processing services for mining and cartographic companies for the last 10 years we may confirm that this confusion appears very common amongst our customers.
Beside inclusive and free software packages which come together with expensive scanners, there are acceptable sets of sophisticated and expensive LiDAR data processing software aimed at various situations: Leica Cyclone, Aveva LFM, Terrasolid Terra Scan, Autodesk Revit, etc. The major problem that plagues these packages is their universality. In other words they may help to resolve various problems with LiDAR data processing but not always be time effective, elegant, convenient and qualitative. Specifically due to their complexity any of these tools require extensive advance training and a high qualification of data analysts.
There is another widely acclaimed myth that LiDAR data is self-sustainable and if one has satisfactory LiDAR results then it is not important to have other types of auxiliary spatial data. Auxiliary data may consist of photos, conventional surveying measurements, seismic cubes or profiles, engineering drawings and 3D models. And it is a common misconception that LiDAR is able to substitute all aforementioned sources.
To some extent this illusion might originate from the fact that high-res LiDAR is extremely voluminous and since it requires tremendous effort to handle millions or even billions of points then it appears unfathomable that other expansive spatial data can be married with LiDAR.
Utilizing significant practical experience with regards to processing any type of LiDAR data we compiled a list of commandments that may appear useful and important for customers. This may apply to those who either want to collect and process LiDAR data or just derive maximum value from LiDAR data already in one’s possession.
1 Clearly formulate final purpose of collection Collection of LiDAR data itself is doubtfully the ultimate goal of any project (some exclusion occur in architectural / artistic applications which we will not touch on in this article). For example if one scans walls of the building in order to calculate number of windows and total window area then at the point of project inception one has to think on how to extract the explicit information related to windows from the point cloud (manually, automatically or semi automatically)and what will be the cost of each scenario.
2 Be aware that buying laser scanning equipment is only half of the job In most cases it costs at least comparable amounts and usually requires much more efforts to acquire specific software packages and proper processing, or both.
3 Be aware that even high quality LiDAR data will not deliver required engineering results Production of engineering data cannot happen without conscious processing of the collected data towards the formulated problem. There are no universal solution and each particular task requires individual analysis. This is why within SightPower’s technology we suggest the 4-tier approach: back-end layer -> generic heavy client (viewer) -> generic metric tools -> special custom tools (custom wizards and instruments). Very often the particular problems can be effectively resolved only at the very top level of this 4-tier structure.
4 Never fully rely on giveaway software which comes together with your scanner In 90% of cases this software won’t possess the necessary features for the resolution of your particular problem
5 Be very careful with buying expensive universal LiDAR data processing packages This statement is right even if the required functionality is explicitly declared. Universal packages rarely provide the most time-effective and the most cost-effective solution of your problem in the long term.
6 Consider combination of LiDAR with any other available types of spatial data Each point cloud is still spatial data and in most cases it may require georeferencing within other data sets. It is worth your deliberation which software can be used for integrating various types of large scale spatial data together in order to leverage the new possibilities which one may obtain by simultaneous usage of various data layers. For instance if one needs to extract construction elements out of some steel structure (power line tower, head gear etc.) one often faces the problem of poor LiDAR point cloud quality for certain parts of construction (shaded by other parts, scanning noise due to various reasons). Taking additional photos of these doubtful areas and integrating these photos with the point cloud scene can drastically improve the quality of post-processing results.
7 Be accurate with LiDAR data filtering Since LiDAR data is voluminous, various software packages suggest to sparse it out in order to make it lighter and so to provide a better manipulation experience. Sometimes filtering works very well and may help to remove some random noise, however in each particular case filtering methods may differ. In other words it is merely impossible to apply generic filters without realizing the purpose of further LiDAR data usage. Most definitely one must not do it just because the available software is not able to handle huge LiDAR data files. The last approach is equal to killing your investment into LiDAR data collection.
8 Organize your LiDAR data Since point clouds are very big then it makes sense to apply specialized methods where one keeps LiDAR metadata in a relational database and actual point cloud files are kept separately in a manageable file system. If LiDAR collection becomes a part of the routine work flow (for instance as a regular method for mine shaft construction surveying support) then organizing point clouds into the manageable databases becomes a critical part of an effective operation.
9 Think about your work flow first If the task is not only to collect and process some sporadic LiDAR data but to make LiDAR data collection and/or processing part of routine operations, one must first think about proper work flow and only then about the tools.
10 Consider the development of custom LiDAR data processing tools Custom tools created for particular needs and in accordance with particular requests usually provide the best solution for any problem. Custom tools may increase the effectiveness and accuracy of a solution by an order of magnitude. As a result one may expect fast return of investment in data collection.
In order to implement the aforementioned recommendations one needs not only to understand their importance but also have the appropriate means and facilities. This is why SightPower offerings have been used in non-trivial complicated situations where the customers couldn’t compromise their risks and reliability.
There are few industrial examples of practical usage of SightPower’s LiDAR data processing technology. The projects mentioned below were fulfilled with the help of SightPower Spatial Enterprise system. Custom processing tools were developed for the resolution of particular problems for each project.
Creation of an engineering model of a mineshaft under construction. (Figures 1 and 6)
Mobile LiDAR scanning technology was used as the source of the point cloud. This project is located in Russia. Scans were produced by DMT GMBH Ltd.
The tasks were
to assess the shift of each tubing ring
to assess the verticality of the whole mineshaft tube
to assess the ellipticity of each tubing ring to construct the database of tubing segments
Construction of this mineshaft has been aggravated by the number of engineering challenges and complex geological and hydrogeologic conditions. This project was peculiar due to the fact that the actual geometry of the mineshaft was assessed not through direct extraction of tubing elements from the point cloud. Instead of that the point cloud was used for accurate geo-reference of project engineering (CAD) model. This approach facilitated the extraction of high-quality surfaces which allowed for filter frosting and so to get the adequate model and actual position of surfaces formally not visible to scanner. Creation of a complex volume vector model out of point cloud generated in the underground mine. (Figures 2 and 3) The project was also in Russia. Construction of precise vector model of connected underground tunnels and chambers was fulfilled with a goal
to provide volume calculations and to check correspondence to the planned project parameters. The following particular problems were resolved: filtering the original scan in order to remove distortions caused by equipment and people located in the mine during scanning session (as well as to exclude other similar artifacts ) classification and separation of surfaces for different tunnels and channels union of separated surfaces to the topologically consistent model with differentiation of notion of "inner" and "outer" volume Creation of headgear models from static laser scans and photos. (Figure 4) The project was done for a South African gold mining company. A static 3D scanning technique was used for initial data collection. The task was to provide detailed engineering models of the headgears and to detect the damaged (twisted, corroded etc.) beams. Each beam had to be specified using the directory of South African steel profiles and sections. In certain poorly scanned areas photos were used in order to advise the software how to extract particular beams and what type of beam must be associated with the particular object.
Creation of anaglyphic images using a combination of airborne LiDAR and orthophotos. (Figure 5)
This project was completed for a Mexican company.
The task was to generate anaglyph images combining airborne LiDAR data and one orthophoto image. This picture gives real "3d-effect" when viewed through anaglyph glasses.
Borys Vorobyov, Ph.D. is the founder and CEO of SightPower Inc. (www.sight-power. com) Borys has 35 years of hands-on experience in spatial data analysis, in particular more than 10 years in LiDAR data processing. Worked for Kharkiv National University in Ukraine, Northwood Technologies, Marconi, and Ambercore Inc. (founder and CTO)
Sergey Reznichenko, Ph.D. is the founder and President of SightPower Inc. Sergey has about 30 years of experience in operational mining and mining innovation development. Worked for RTZ, De Beers (Head of DebTech department), Ambercore Inc. (Vice-President of Mining).
Vitalii Monastyrov, M.Sc. is the head of custom project department at SightPower Inc. Vitalii has about 15 years of spatial data analysis experience, in particular about 9 years in LiDAR data processing. Worked for SPAERO (Kharkiv, Ukraine) and then for Ambercore Inc.
A 2.942Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE