himhan34 / imreg_fmt_odometry Goto Github PK

imreg_fmt is an implementation of the Fourier-Mellin transform-based image registration method originally described by Reddy and Chatterji [1]. Given a pair of images, the algorithm computes the translation (x, y), scale and rotation required to register/align one image to the other. A brief explanation of the algorithm is provided here.

This project is a partial port of the Python implementation by Christoph Gohlke and Matěj Týč (see here). It is written in C++ and is suited for registering a sequence of images (such as from a video). For images of size 320x240, the algorithm runs at approximately 14 Hz on an Intel Core i3 (1.7 GHz).

• fftw3
• OpenCV 4.2
• Eigen

mkdir build && cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make

• You might have to add the following flag when running cmake (see here)

  cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_FLAGS='-isystem /usr/local/include' ..

./image_main <path to first image> <path to second image>
./video_main <path to video>

This project is licensed under the GPLv3 License. The license of the original Python version by Gohlke and Týč can be found here.

[1] B. S. Reddy and B. N. Chatterji, “An FFT-based Technique for Translation, Rotation, and Scale-Invariant Image Registration,” IEEE Transactions on Image Processing, vol. 5, no. 8, pp. 1266–1271, 1996.