semi-global-matching

open Structure-fom-Motion and Multi-View Stereo reconstruction library

SGM，立体匹配StereoMatching最经典应用最广泛算法，4000+引用，兼顾效率和效果。完整实现，代码规范，注释清晰，博客教学！

Stereo-image depth reconstruction with different matching costs and matching algorithms in Python using Numpy and Numba

depth map computation

End-to-end stereo depth estimation workflow using Semi-Global Matching (SGM), bridging Python reference implementations and FPGA-oriented hardware designs with transparent validation and visual comparison.

Optimized (very fast) stereo matching algorithms in MATLAB and Python. It includes implementations of Block Matching, Dynamic Programming, Semi-Global Matching and Belief Propagation.

University course

Stereo disparity estimation by Normalized Cross Correlation, SGM algorithms, and performance optimization.

A C++ implementation of Semi-Global Matching (SGM) for stereo matching. It uses the two-pass 8-directions version of the algorithm with a small improvement for better results.

MATLAB implementations of the Semi-Global Matching (SGM) stereo matching algorithm, featuring 4-path, 8-path and 16-path cost aggregation variants for dense disparity map estimation.

Semi-global matching for Rust

Evolution of Basic Stereo Matching Algorithms. From Block Matching and Dynamic Programming to Semi-Global Matching and Belief Propagation. Compare the source code to see the improvements.

Foundation4D (arXiv 2610.30340) in dependency-free Rust: factorized spatiotemporal attention, SGM plane-sweep depth, occupancy, 3D tracking, pose-graph optimization and WebGPU Gaussian splatting — running end-to-end in the browser via WASM.

Honest stereo disparity benchmarking: masked metrics, exact synthetic ground truth with occlusion, and accuracy-vs-latency Pareto frontiers. Runs with no GPU.