homography-decomp

C++ Implementation of Homography Decomposition

This is implementation of homography decomposition to rotation(s), translation(s) and plane normals. The implementation is based on the paper by Ezio Malis, et.al. "Deeper understanding of the homography decomposition for vision-based control". 2007

Usage: Create HomographyDecomp object and call the method

   void decomposeHomography(const cv::Matx33d& H, const cv::Matx33d& K,
                                 std::vector<CameraMotion>& camMotions);

Inputs:

H    - Homography matrix between two images (3x3 )
K    - Intrinsic camera matrix. (3x3)

Outputs:

camMotions - Array of {R, t, n}. R-rotation matrix, t-translation vector, n-plane normal vector 

This implementation is integrated with OpenCV library. See the documenation in OpenCV - decomposeHomographyMat()