Colmap

COLMAP is a 3D scene reconstruction suite and C++ geometry library that implements a full structure-from-motion pipeline. It functions as a GPU-accelerated photogrammetry tool and multi-view stereo framework designed to produce dense 3D geometry and watertight meshes from collections of 2D images.

The project distinguishes itself through hardware-accelerated feature extraction and a modular camera modeling system that supports perspective, fisheye, and equirectangular lens types. It employs vocabulary tree image retrieval to efficiently identify similar images in large datasets and provides Python language bindings to integrate its core reconstruction engine into custom applications.

The software covers a broad range of photogrammetry capabilities, including camera pose estimation, bundle adjustment for reprojection error minimization, and image undistortion. Its pipeline extends from initial visual feature matching and geometric estimation to dense 3D model generation, surface reconstruction, and mesh texturing.

Users can manage the reconstruction process and visualize spatial data through a graphical user interface or automate workflows via a command-line interface.

Features

3D Reconstruction - Provides a comprehensive suite for extracting image features, performing bundle adjustment, and texturing 3D models.
Structure from Motion Pipelines - Generates 3D models from collections of 2D images using a full structure-from-motion pipeline.
Multi-View Depth Estimators - Computes depth maps from multiple calibrated views to generate dense 3D point clouds.
SIFT Feature Accelerators - Leverages CUDA-capable graphics cards to accelerate the extraction of visual landmarks from images.
Multi-View Stereo Generators - Produces detailed 3D geometry from multiple views using multi-view stereo processing.
Watertight Mesh Generation - Converts sparse point clouds into watertight 3D surfaces and textured meshes for external visualization.
Pose Estimation - Implements algorithms to recover exact camera positions and orientations from image features and GPS data.
Camera Calibration - Determines intrinsic and extrinsic camera parameters to handle lens distortions across various camera models.
Multi-View Scene Reconstructions - Produces dense 3D geometry and watertight meshes from multiple calibrated images of a scene.
Structure From Motion - Reconstructs 3D scenes from 2D image collections using camera pose estimation and point cloud generation.
Feature Extraction - Detects and describes local image features using GPU-accelerated algorithms to identify unique visual landmarks.
Local Feature Matching - Identifies corresponding visual landmarks across large image datasets to establish geometric relationships.
3D Geometry Engines - Implements a high-performance C++ engine for processing 3D geometric data and transformations with Python bindings.
GPU-Accelerated Photogrammetry - Uses hardware acceleration to perform high-speed visual landmark detection and image feature matching.
Photogrammetry Orchestrators - Provides a command-line toolkit that orchestrates the entire photogrammetry pipeline from feature extraction to texturing.
3D Reconstruction Pipelines - Provides an automated pipeline for converting sets of 2D photographs into accurate 3D digital models.
Bundle Adjustment Algorithms - Implements optimization techniques to refine 3D coordinates and camera poses by minimizing reprojection errors.
Camera Geometry Estimation - Calculates absolute poses, matrices, and homographies using robust estimators to determine spatial relationships.
Multi-Lens Projection Models - Implements interchangeable mathematical models to handle different lens types like fisheye or equirectangular.
Image-to-Image Retrieval - Finds visually similar images within large datasets using vocabulary trees and embedding techniques.
Large-Scale Image Retrieval - Indexes visual features in a hierarchical vocabulary tree to efficiently find similar images in large datasets.
Point Cloud Surface Extraction - Generates 3D meshes from point clouds using surface reconstruction algorithms and visibility-based filtering.
Database Import Pipelines - Extracts and matches visual features from images and imports them into a database for geometric verification.
Lens Distortion Removal - Removes lens distortion from images based on camera models to produce rectified views.
3D Point Cloud Filtering - Provides algorithms for removing noise and outliers from 3D point cloud datasets to clean sparse models.
Photogrammetric Texture Mapping - Produces textured 3D models by mapping colors from calibrated original images onto the reconstructed mesh.
Watertight Mesh Generation - Creates watertight surfaces from point clouds using meshing and simplification algorithms to refine geometry.
Python-C++ Bindings - Provides Python-C++ bindings that expose high-performance reconstruction functions to Python scripts.
Geospatial Model Alignments - Aligns reconstructed 3D models to global coordinate systems using GPS data and gravity priors.
Photogrammetry and Scanning - Automatic 3D reconstruction from image collections.
Structure From Motion - Comprehensive pipeline for structure from motion and multi-view stereo.
Media & Communication - Structure-from-motion and multi-view stereo

cdcseacave/openMVS

4,021View on GitHub

openMVS is a multi-view stereo library and photogrammetry pipeline used for 3D scene reconstruction. It transforms Structure from Motion data—specifically camera poses and sparse point clouds—into detailed 3D models consisting of dense point clouds and textured meshes. The project provides a sequence of processing stages to densify point clouds, generate 3D surface meshes, and apply photorealistic textures. It uses multi-view texture blending to map accurate colors onto reconstructed geometry and employs iterative refinement to optimize mesh details. The system includes capabilities for impo

openMVG/openMVG

6,451View on GitHub

openMVG is a computer vision geometry library and toolkit for multiple view geometry. It serves as a framework for structure from motion and 3D scene reconstruction, providing the tools necessary to recover 3D point clouds and camera poses from collections of 2D images. The library implements both global and incremental structure-from-motion pipelines. It uses geometric algorithms to calculate camera pose estimation and image localization, employing Levenberg-Marquardt bundle adjustment to refine 3D coordinates and camera parameters by minimizing reprojection error. The project covers a broa

OpenDroneMap/ODM

5,853View on GitHub

OpenDroneMap (ODM) is an open-source aerial drone photogrammetry pipeline that converts 2D images into georeferenced 3D models, orthophotos, point clouds, and digital elevation maps. At its core, the OpenDroneMap Processing Engine orchestrates a complete Structure-from-Motion workflow, from feature extraction through dense reconstruction and tiled output generation, purpose-built for transforming drone-captured imagery into geospatial data products. The toolkit distinguishes itself through GPU-accelerated SIFT feature extraction using CUDA-capable NVIDIA graphics cards, roughly doubling proce

cvg/Hierarchical-Localization

3,961View on GitHub

This project is a 3D visual localization framework designed to determine a camera's exact position and orientation by matching 2D image features against a 3D reference model. It includes a structure-from-motion pipeline to reconstruct 3D scene geometry from unordered image sets, creating the necessary spatial maps for localization. The system employs a hierarchical coarse-to-fine localization approach. This process begins with a global-descriptor image retrieval system to identify candidate reference images from a large database and progresses through local feature matching to final 3D model

colmapcolmap

Features