The Ongoing Debate

There always seems to be a running debate regarding drone mapping with LIDAR versus drone mapping with photogrammetry.

LIDAR requires a single laser pulse return from object space (e.g. tree, ground, building, etc.) to the sensor. Photogrammetry (these days always “multi-ray”) requires that each spot in object space that is to appear in the photogrammetry model must be “seen” by multiple images (usually 5+) that “see” the same ground point, but from differing camera perspectives. This, of course, is what makes LIDAR the only possible solution in occluded areas such as regions with trees (vegetation “canopy”).

True View 3D Imaging Sensors (3DIS) provides both LIDAR and integrated dual cameras giving users the ability to use either. However, a 3DIS in fusion mode gives you the best of both worlds; the single ray penetration capability of LIDAR with the colorization from the Red-Green-Blue (RGB) cameras.

Thus, this debate does not come up with a True View 3DIS (a True View 410, 615 or 620). Therefore, the question really becomes “Can I get away with just photogrammetry because it is a much less expensive way to enter the drone mapping business!”

What Are My Limitations?

This is actually pretty easy to answer, I think. 

• If you have a True View 3DIS, you can handle any drone mapping job. 
• If you have a LIDAR only, then you can address LIDAR projects, but not projects that require delivery of color products such as a colorized point cloud (3D Imagery) or orthomosaics.
• If you have photogrammetry, you will not be able to address projects with overhead cover (e.g. a topo project in a vegetated area), wire extraction (e.g. transmission lines) or projects that require modeling thin overhead structures such as piping or conveyors. 

LIDAR vs Photogrammetry: Real World Example

Let’s pause for a moment to look at a section of a real world data set. We are doing a bit of R&D on earthen dams. These dams in Alabama are usually in heavily vegetated areas where photogrammetry is not an option. 

Figure 1 shows an area of a dam site where an access road runs through a vegetated section. I have indicated a few of the “void” problems using superimposed red arrows. Note in the profile view (lower section of the image), the road surface is apparent, but there are no ground shots at all on either side of the road.  

Also, notice the trees are amorphous blobs rather than well defined structures. Another problem is in the ortho coverage itself. In the upper section of the view (plan view), notice the large areas where there is simply no data at all.

This occurs because the photogrammetry software was not able to achieve a successful “correlation” (matches) between equivalent spots in the overlapping images. Without sufficient matches, the software has no choice but declaring a void. Obviously, you could not make a stakeholder happy with such a data set.