How exactly does PIX4D Mapper produce a photogrammetric point cloud from UAV photos?

I am using the photogrammetric point clouds, DTM and DSM produced in PIX4D Mapper as part of PhD research. It would be very helpful, and very fascinating, to know how exactly these point clouds, DSM and DTM are produced just from UAV photos. I assume the exact algorithms are patented and PIX4D doesn’t publish them in detail, but can anyone tell me more about the underlying processes/algorithms used to produce them?

PIX4Dmapper utilizes advanced photogrammetric techniques to generate point clouds, Digital Surface Models DSM , and Digital Terrain Models DTM from UAV photos. Here’s a general overview of the processes involved:

  1. Image Acquisition : UAVs capture overlapping images of the area of interest. These images are geotagged with GPS data to provide spatial context.

  2. Image Alignment : The software identifies common features across multiple images to align them. This process involves detecting key points and matching them across images to establish a coherent structure.

  3. Point Cloud Generation : Using the aligned images, PIX4Dmapper generates a dense point cloud. This involves triangulating the matched key points to create a 3D representation of the area.

  4. DSM Creation : The point cloud is used to create a DSM, which represents the surface of the area, including all objects like buildings and vegetation.

  5. DTM Generation : To produce a DTM, the software applies classification algorithms to the DSM to remove non-ground objects, resulting in a model that represents the bare earth terrain.