A 2.933Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
The mapping process has benefited tremendously from the latest advances in sensors technologies, computer processing power, and computer vision-based processing algorithms. Certain benefits include the prospect of transitioning from a two-dimensional map environment to a three-dimensional one. Up until recently, the two-dimensional representation of the terrain features is used to evaluate ground contents, and was the only method used by most GIS enterprises. Even in the evaluation of the 3D terrain topography, we often use two-dimensional representation such as contours or shaded relief. Such representation of the terrain deprives users from appreciating the three-dimensional nature of the ground features and from what it entails by limiting the mapping process. Why don’t we then utilize three-dimensional GIS instead of the two-dimensional one we are using now? To address such a question, there are many reasons or obstacles preventing us from efficiently integrating three-dimensional GIS (3D GIS) into our daily mapping activities, some of these challenges include the following:
1. The High Cost of High Resolution Terrain Data: True 3D GIS is only achievable when the smallest image element, or pixel, is associated with a unique ground (object) elevation derived from high definition Digital Surface Model (DSM). Due to its high production costs, the generation of high quality and dense DSM is prohibitive. Current LiDAR technologies prevented us from producing DSM with resolution that matched the high resolution of imagery. In addition, past performance of auto-correlation and image matching algorithms failed to convince users that high quality terrain elevation data could be produced from imagery with a resolution identical to the resolution of the imagery, i.e. pixel level resolution.
2. Limitations in the ThreeDimensional Viewing Technology: Users of geospatial data oftentimes have limited options when it comes to viewing data and performing different mapping activities in 3D environment. Such 3D mapping activities are only performed today with the help of stereo viewing hardware and glasses. GIS data users need to be able to view and work with 3D GIS data using standard computer hardware configuration.
Recent achievements in DSM generation, though, are changing the pace of our transition to true 3D GIS. Based on what was presented during the ILMF2015 conference in Denver, we are definitely witnessing the right signs and a turning point towards the efficient integration of 3D GIS. Among the signs that mark the birth of 3D GIS include the following:
1. The Recent Offering of Geiger Mode and Photon Counting LiDAR Services: The focal plane based LiDAR design will reduce the cost of high resolution LiDAR data. In an ILMF2015 session focused on Geiger mode and photon counting LiDAR, speakers from companies such as Sigma Space, 3DEO, Princeton Light Wave, and HARRIS offered commercial solutions for a dense LiDAR collection based on focal plane-based design LiDAR systems. Some of the new LiDAR systems presented in the session were able to collect data with sampling rate of more than 200 MHz, versus the maximum 800 KHz of the current linear LiDAR, and point cloud density of up to 100 points per square meter. Such LiDAR systems collect data from an altitude up to 30,000 ft. above ground resulting in wider ground coverage. The most important aspect that these new LiDAR technologies offer, beside the density, is the focal plane design aspect which results in a raster-style data acquisition. Such mode is important in providing an elevation for every single pixel of the multi-spectral imaging that will be discussed next. Having an elevation for every single pixel of the scene is an important milestone to achieve toward 3D GIS.
2. The Multi-spectral LiDAR: Another sign that 3D GIS emerging is the latest offering of multi-spectral LiDAR that collects elevation data with multi-spectral characteristics. The ILMF2015 sessions also included presentations on the first commercial offering of such LiDAR. The new LiDAR, Titan by Optech, uses three laser beams selected at three different regions of the electromagnetic spectrum spanning the visible and the near infrared. Such LiDAR system will be very effective when the laser technology and perhaps physics enable manufacturers of such LiDAR to use all the correct four bands from the electromagnetic spectrum namely, red (610660 nm), green (530580 nm), blue (435495 nm), and near infrared (8401,000 nm) that are required to produce natural colors (RGB) and colored infrared (CIR) imagery beside topographic and shallow bathymetric elevation data. The introduction of Titan is just the beginning of more advanced and perhaps more capable multispectral LiDAR systems. Shallow bathymetric and multi-spectral mapping capabilities of such system are demonstrated in Figures 1 and 2. I expect other leading LiDAR manufacturers to offer their own versions of such multi-spectral LiDAR systems in the near future.
3. The Recent Image Correlation Capabilities: Until recently, users never trusted the quality of a digital surface model derived from stereo image matching and auto-correlation. Recent introduction of more sophisticated and more accurate images matching algorithms and techniques such as Semi Global Matching (SGM), Scale-invariant Feature Transform (SIFT) and Structure from Motion (SfM), in building a three-dimensional surface model from imagery are taking us a step closer to 3D GIS. Software such Smart3DCapture from Acute3D (recently acquired by Bentley Systems) and PhotoMesh by Skyline Software Systems Inc. offer mature solutions for the production of high definition DSM from imagery. I was recently involved with a project within Woolpert involving three-dimensional modeling of the Devil Tower landmark in Wyoming using Acute3D software. We flew the site from multiple views and multiple look angles resulting in imagery with a ground sampling distance of 7.5 cm. Figure 3 shows the resulting 3D textured model generated for that landmark using SmartCapture of Acute3D.
Figure 4 represents realistic 3D model with 2 cm. resolution for part of the city of Frederick, Maryland using Skyline’s PhotoMesh.
The advantage of producing a LiDAR-like or better quality DSM from imagery is huge to the project budget as it in many cases eliminates the needs for acquiring LiDAR data for the project area. Such new capability requires new standard for image planning and acquisition that are closely aligned with the image acquisition required for oblique imagery. Multi-view imaging is essential for the successful generation of accurate DSM using the new algorithms.
Finally, we are definitely witnessing great signs for the full implementation of true 3D GIS where a pixel in space is defined by its picture element (spectral characteristics) and elevation to form a 3D pixel. Sooner or later, the three dimensional modeling environment will become the normal way of serving geospatial data. In such environment every pixel in the scene has an elevation associated with it and both pixel spectral characteristics and elevation information are stored in the same file on a regular basis. With the full 3D GIS implementation, future aerial data acquisition sensors are expected to operate as multi-spectral LiDAR system (or 3D camera) that collects four (or more) co-registered products such as RGB imagery, CIR imagery, topographic LiDAR data, and shallow bathy LiDAR data. Such system will most likely be based on photon counting or Geiger mode LiDAR technology with multi-spectral capability. When such system exists, supporting technologies and data standards such as raster file format will follow the progress achieved in data acquisition technology to provide the fertile ground and the infrastructure necessary for the full implementation of 3D GIS.
Dr. Qassim Abdullah is a Senior Geospatial Scientist and associate at Woolpert, Inc. Dr. Abdullah publishes the monthly column "Mapping Matters", in the ASPRS journal PE&RS and he is the recipient of the 2010 prestigious Photogrammetric Fairchild award of the ASPRS.
A 2.933Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE