Structure from Motion
The photogrammetric algorithm that reconstructs 3D geometry from a set of overlapping 2D photographs by tracking common feature points across images.
Structure from Motion (SfM) is the computational process underlying modern close-range photogrammetry. It takes a set of overlapping photographs — taken from different positions around a subject — and identifies common feature points across multiple images. From the known position and angle of each camera relative to these matched points, it simultaneously solves for both the 3D structure of the scene and the positions of the cameras.
The name comes from the insight that a camera moving around a static scene can reconstruct its 3D structure — the structure comes from the motion. SfM is what makes consumer cameras useful for precise 3D documentation: the photographer moves around the subject capturing photographs, and the algorithm reconstructs the geometry from that motion.
In practice, SfM processing is performed in photogrammetry software after image capture. The key parameters the operator controls are the density of the computed point cloud and the accuracy target for the alignment step. Ground Control Points (GCPs) are used to scale and georeference the model when spatial accuracy is required.