Photogrammetry is Hard
Photogrammetry is the science of combining 2D photos into 3D models and data, and it is exactly that: a science. The entire amount of knowledge required to be a professional photogrammetrist requires an entire career with years of study, and is far beyond the scope of this initial primer. While advanced software and online processing tools have made the process of photogrammetry considerably easier, Aerotas still strongly recommends that photogrammetry for surveying and engineering applications be performed by someone trained in the science, and certified by the American Society for Photogrammetry and Remote Sensing (ASPRS).
The first step in photogrammetry is the initial processing of the images. In this step, a computer algorithm creates a set of “key points” in each image that are used to match images to one another. Then, the computer will compare these key points to one another and figure out which images are next to one another. Using these matches, the computer can also automatically calibrate the camera’s lens parameters by seeing how the key points were matched to one another. Using these matches, the software can create a sparse point cloud of key points that approximates the surface of the model, but at low fidelity and accuracy.
Ground control integration can happen either before or after initial processing. It is typically easier to spot the GCP across multiple photos if performed after initial processing, but that does take more processing time. The purpose of GCP integration is to anchor the model to know coordinates on the ground, and ensure that the coordinate system used in ground surveying properly matches that which is used in the aerial survey.