This companion implements Otsu thresholding from scratch and compares its threshold and binary output with OpenCV. Both Python and C++ paths are headless, validate their inputs, and resolve default assets relative to this directory.

• Python 3.10–3.14 with OpenCV 4.13 or newer
• CMake 3.16 or newer
• A C++17 compiler
• OpenCV 4 or OpenCV 5 development libraries for the C++ examples

Install the Python dependencies:

python -m pip install -r requirements.txt

Run the custom implementation and the OpenCV reference:

python otsu_implementation.py
python otsu_method.py

Both commands write results under outputs/. Use --help to select another input or output location. No display server is required.

Run the deterministic regression tests:

python -m pytest -q tests/test_otsu.py

cmake -S . -B build -DBUILD_TESTING=ON
cmake --build build --config Release
ctest --test-dir build --output-on-failure -C Release
./build/otsu_implementation
./build/otsu_method

The regression test checks the real boat.jpg fixture, synthetic bimodal data, constant images, invalid inputs, and pixel-for-pixel parity with OpenCV. For boat.jpg, both implementations select threshold 132.

LearnOpenCV is maintained by BigVision.AI, a computer vision and AI consulting company. We help organizations design, build, optimize, and deploy production-ready AI solutions. Our team has deep expertise in computer vision, deep learning, multimodal AI, and edge deployment, with experience solving complex technical challenges across industries.

Have a project in mind? Talk with our expert AI solution builders.