3D Reconstruction is the computational process of recovering three-dimensional geometric and structural information from multiple 2D images or sensor data (such as LiDAR or depth cameras) using techniques including Computer Vision, photogrammetry, and Structure-from-Motion (SfM), enabling digital capture of real-world objects and environments for Digital Twin creation, immersive environment mapping, and spatial analysis.
Semantic Classification
Content
Overview
3D Reconstruction bridges the physical and digital worlds by algorithmically deriving 3D structure from 2D observations. Key methodologies include Structure-from-Motion (recovering both structure and camera motion from video), multi-view stereo (dense depth estimation), and sensor fusion combining multiple data streams.
Primary Techniques
-
Structure-from-Motion: Extracting 3D geometry from overlapping photographs and calculating camera trajectories
-
Multi-View Stereo (MVS): Dense depth estimation by analysing matching pixels across multiple images
-
Photogrammetry: Professional 3D capture using calibrated imaging workflows
-
LiDAR Scanning: Direct depth measurement using laser time-of-flight sensors
-
Depth Sensors: Real-time 3D acquisition via structured light or time-of-flight cameras
Applications
-
Heritage Digitisation: Preserving cultural artefacts and archaeological sites
-
Architectural Scanning: Creating as-built models of buildings
-
Industrial Inspection: Quality control through precise dimensional analysis
-
Real Estate: Virtual property tours via captured environments
Related Concepts
-
Computer Vision, Photogrammetry, Point Cloud, Digital Twin, Structure-from-Motion