I canβt specify any particular book right now, but depending on your mathematical background, I would suggest going in that order
- Vector and linear algebra, intermediate level, up to matrix operations, LU decomposition, cross product.
- Projective geometry, to homogeneous coordinates, plane homography
- 3D graphics, viewing and projecting matrix, frustum
- Basics of image processing, thresholds, edge detection, line detection.
After these four two you can understand the tracking of rectangular markers
- Calculus of many variables, Fourier transform, DFT
- Least square method
- Intermediate linear algebra, eigenvalues, eigenvectors, SVD
- Extended numerical methods, nonlinear least squares, Gauss-Newton, Levenberg-Marquardt
- Advanced image processing, blob detection SIFT / SURF / FAST
- Intermediate projective geometry: basic and fundamental matrices, epipolar geometry
- Kit adjustment
After that you can understand mirrorless tracking
And even more advanced math, which is used in the cutting edge AR:
- Understanding the basics of Lie groups and algebras
- Statistics, reliable estimates
- Quaternions
- Kalman Filters
- Clifford Algebras (Geometric Algebra) - A Generalization of Quaternions
- Bursts
- Extended projective geometry (e.g. trifocal tensor, 5-point algorithm)
mirror2image
source share