Structure from Motion (SfM)

Structure from Motion (SfM) is a computer visionComputer VisionComputer Vision is a field of artificial intelligence that enables machines to interpret and understand visual inform... and photogrammetric technique that estimates three-dimensional structure from a series of overlapping two-dimensional images captured from multiple viewpoints. Unlike traditional photogrammetryPhotogrammetryPhotogrammetry is the science of extracting precise measurements and creating 3D models from photographs. It is widel... that requires precisely known camera positions and calibration, SfM simultaneously solves for both the 3D scene geometry and the camera positions and orientations, making it highly flexible and accessible. How SfM worksThe SfM pipeline begins with feature detection (identifying distinctive keypoints in each image), followed by feature matching (finding corresponding points across image pairs), bundle adjustment (simultaneously optimizing all camera positions and 3D point coordinates), and dense point cloudPoint CloudA point cloud is a large set of three-dimensional data points representing the external surface of objects or terrain... generation (computing 3D coordinates for millions of surface points). The result is a dense, georeferenced 3D point cloud that can be converted to meshes, digital elevation models, and orthomosaics. ApplicationsSfM is the core technology behind modern drone mappingDrone MappingDrone mapping uses unmanned aerial vehicles (UAVs) equipped with cameras and sensors to capture high-resolution aeria... software (Pix4D, Agisoft Metashape, OpenDroneMap), enabling rapid generation of survey-grade products from consumer-grade cameras. It is used in geological outcrop mapping, archaeological site documentation, glacial monitoring, forensic scene reconstruction, and cultural heritage preservation. SfM has democratized 3D mapping3D Mapping3D mapping leverages technologies like LiDAR and photogrammetry to produce detailed three-dimensional representations... by enabling accurate three-dimensional reconstruction from ordinary photographs. Its flexibility, low equipment requirements, and integration with drone platforms have made it one of the most widely adopted 3D data collection methods in geospatial science.