Optical Flow - Lucas-Kanade Algorithm

This repository contains a small C++/OpenCV implementation of sparse optical flow using the Lucas-Kanade method.

The program compares two consecutive sample images:

• test/data/img/scene00140.png
• test/data/img/scene00141.png

It estimates local motion vectors and draws them on the first frame.

Requirements For The macOS Build

• C++17 compiler
• Xcode Command Line Tools

The macOS build does not require OpenCV, CMake, or Homebrew.

Build On macOS Without OpenCV Or CMake

The repository includes a macOS-native implementation that uses Apple's built-in ImageIO/CoreGraphics frameworks for PNG input/output. It does not require Homebrew, OpenCV, or CMake.

Build:

make mac

Run:

make run-mac

The result is saved as:

output_mac.png

Build With CMake

This path is optional and requires OpenCV 4.x plus CMake 3.16 or newer.

From the repository root:

cmake -S . -B build
cmake --build build

Run:

./build/lucas_kanade_optical_flow

On Windows, the executable may be under a configuration directory:

.\build\Debug\lucas_kanade_optical_flow.exe

Build With Visual Studio

Open Opencv_test.sln in Visual Studio and build the x64 configuration.

The Visual Studio project expects OpenCV at:

C:\opencv\build

If OpenCV is installed somewhere else, define the OpenCVRoot environment variable before opening Visual Studio:

setx OpenCVRoot "C:\path\to\opencv\build"

The project currently links against OpenCV 4.7.0 world libraries:

• Debug: opencv_world470d.lib
• Release: opencv_world470.lib

If your OpenCV version is different, update the library names in test/test.vcxproj.

Notes

The implementation follows the Lucas-Kanade least-squares form:

[Ix Iy] [u v]^T = -It

Low-texture windows are skipped using a minimum eigenvalue threshold to avoid unstable motion vectors.