Structure-from-Motion (SfM)
Definition
A photogrammetric technique in which overlapping images from a drone or camera, processed by software, generate a three-dimensional point cloud and textured mesh model of the scene. Used for fire scene and post-blast scene documentation, producing a dimensionally accurate spatial record that survives demolition.
- Output
- 3D point cloud and textured mesh model
- Key forensic use
- Fire scene and post-blast documentation
- Core method
- Feature matching across overlapping images to estimate camera positions and scene structure
- Scaling tool
- Ground control points (GCPs) with known coordinates for accurate measurement and georeferencing
Common questions
How does Structure from Motion work without a pre-calibrated camera?+
SfM estimates camera positions and 3D geometry simultaneously by matching feature points (corners, edges, distinctive patches) across overlapping photographs. Software solves for both the camera poses and scene structure at once, so you don't need to know the camera calibration beforehand. This makes it practical for any overlapping photos.
Why is SfM useful for fire and blast scenes?+
SfM creates a dimensionally accurate spatial record (point cloud and textured mesh) that survives demolition. Once the scene is gone, you still have a measurable 3D model. Ground control points with known coordinates can scale and georeference the model so measurements are trustworthy.
What hardware do you need to capture the images?+
A drone or hand-held camera, depending on the scene. You need overlapping photographs of the area. No dedicated 3D scanner hardware is required. The software handles all the 3D reconstruction from the 2D images.
Related terms
- Terrestrial Laser Scanning (TLS)
- A ground-based 3D LIDAR survey instrument that captures millions of range measurements per scan to produce a millimetre-accurate point cloud of the...
- Agisoft Metashape
- Commercial photogrammetry software (Agisoft LLC, Saint Petersburg) widely used in forensic anthropology for 3D model production from image sets. Accuracy of approximately...
- Cal Fire (California Department of Forestry and Fire Protection)
- California's state fire agency, which has integrated drone aerial survey and multispectral imaging into major wildfire investigations since 2018. Operational use in...
- Camera Calibration
- The process of determining the camera's intrinsic parameters: focal length, principal point, and lens distortion coefficients. Calibration can be done from a...
- COLMAP
- Open-source SfM photogrammetry software maintained by ETH Zurich, providing zero-cost 3D reconstruction from image sets. Used in under-resourced forensic laboratories in South...
- DBSML
- Detroit-Bone Sex/Maturity Logic: a logistic-regression-based classification system for sex estimation from cranial and postcranial linear measurements, developed at Wayne State University, providing...
- Dense Point Cloud
- A set of millions of three-dimensional coordinate points, each with an RGB colour derived from the photographs, that together represent the surface...
- ENFSI EWG AI Working Group
- The European Network of Forensic Science Institutes Explosives Working Group's artificial intelligence working group, active from 2021. Produced a 2023 position paper...
- Explainability (ML Forensic)
- The capacity of a machine-learning model to articulate, in interpretable terms, the features driving its classification or estimation output. Required under Daubert,...
- Geomorph (R Package)
- An open-source R package (Adams and Otarola-Castillo 2013; actively maintained through 2024) implementing Procrustes superimposition, principal component analysis in shape space, multivariate...
- Ground Control Point (GCP)
- A physical marker in the scene whose three-dimensional coordinates have been measured independently by total station or GPS. Assigning those coordinates to...
- Handheld Raman
- A portable spectrometer using a near-infrared laser (typically 785 nm or 1064 nm) to measure inelastic light scattering from molecular bonds. Generates...
Explained in these topics
- Close-Range Photogrammetry and 3D RecordingA photogrammetric computation method that estimates three-dimensional structure and camera pose simultaneously from a set of overlapping images, using feature...
- Emerging Methods: 3D Scanning and Geometric MorphometricsThe computational framework underlying photogrammetry: camera positions are estimated from matching feature points across overlapping photographs, enabling 3D...
- Emerging Methods: Handheld Raman, LIBS, CT Imaging, MLA photogrammetric technique in which overlapping images from a drone or camera, processed by software, generate a three-dimensional point cloud and textured me...
- Photogrammetry and Metric Analysis from ImagesA multi-image technique that simultaneously estimates camera positions and scene geometry from overlapping photographs, producing a measurable 3D point cloud....