COLMAP

COLMAP is a general-purpose pipeline for 3D reconstruction from unordered image collections, specializing in Structure-from-Motion (SfM) and Multi-View Stereo (MVS). It is an open-source project designed to produce sparse camera poses and dense 3D reconstructions, with support for texturing. COLMAP emphasizes automation and robustness, offering end-to-end processing from raw images to final models.

The software provides a complete SfM toolchain and an accompanying dense-reconstruction module. In SfM, COLMAP performs feature detection and description (commonly SIFT), image-to-image matching with geometric verification, and camera pose estimation. It supports both incremental and global SfM strategies, followed by bundle adjustment to optimize camera parameters and 3D points. The MVS component computes dense depth maps for multiple views and fuses them into a dense point cloud, which can be enhanced into a mesh with optional texture mapping.

COLMAP is built as a modular, cross-platform tool with both a graphical user interface and a command-line interface. It is written in C++ and relies on common computer-vision and optimization libraries, including OpenCV, Eigen, and the Ceres Solver, with Qt-based UI and optional GPU acceleration. Output formats typically include standard 3D data files such as dense point clouds, meshes, and textured models (e.g., PLY, OBJ), along with camera parameters and sparse reconstructions. COLMAP is widely used in academic and applied research for tasks ranging from cultural heritage documentation to robotics and augmented reality. It is released under the GNU General Public License.