approachesphotogrammetry

Photogrammetry is the science of making measurements from photographs, especially to recover precise 3D coordinates and other spatial information. Approaches in photogrammetry encompass traditional methods based on stereo imagery as well as modern computational workflows that operate on large sets of images. They vary by data source, scale, and processing goals, but share the aim of deriving accurate geometry from visual data.

Traditional and close-range approaches rely on calibrated cameras and overlapping imagery. In stereo photogrammetry, two or more photographs with known relative positions are used to triangulate 3D points through parallax, enabling metric surveys, mapping, and modeling of terrain and objects. Terrestrial or close-range photogrammetry places cameras near the object, often using structured grids or hand-held photography, to capture fine detail for cultural heritage, industrial inspection, or architectural documentation.

Structure-from-Motion and multi-view stereo are dominant modern approaches. SfM estimates camera positions and a sparse 3D structure from unordered image sets without prior knowledge, then MVS densifies the scene to produce dense point clouds, meshes, and textures. These pipelines can operate autonomously or with ground control to improve scale and georeferencing.

Processing typically includes camera calibration to model lens distortion, feature-based matching to establish correspondences, and bundle adjustment to optimize camera parameters and 3D points. Outputs include orthophotos, digital elevation models, point clouds, meshes, and textured models. Accuracy depends on image quality, overlap, geometry, calibration, and the use of control points or known references.

Applications span mapping, surveying, archaeology, architecture, forestry, and film production. While photogrammetry is powerful for non-contact measurement, challenges include occlusions, reflective or transparent surfaces, repetitive textures, and the need for substantial computing resources for dense reconstructions.