Dat Vu add
Use Octave octave RunMeBatch.m
VLfeat: https://drive.google.com/drive/folders/1jV5RR3vDg700-ifMU0aVSxdqj8c6vHR7?usp=sharing
issues: vlfeat/vlfeat#188 and https://stackoverflow.com/questions/25209505/vlfeat-in-octave-invalid-use-of-script-in-index-expression
Epipolar Rectification Toolkit
by Andrea Fusiello, 2018
This toolkit is a collection of Matlab function implementing epipolar rectification both in the calibrated and uncalibrated case. It requires some functions contained in the Computer Vision Toolkit (http://www.diegm.uniud.it/fusiello/sw/ComputerVisionToolkit.zip) by the same author and the VLfeat toolbox (http://www.vlfeat.org) by A. Vedaldi, for the SIFT implementation.
After installing the abovementioned dependencies, cd to the toolkit main directory and type runMe. This will call the calibrated rectification and the uncalibrated one, in a sequence.
The data is in the "examples" folder. It should be easy to replace it
with your own images. The 3x4 camera matrix, or PPM, is contained in a
text file with extension ".pm".
While in the calibrated case results are always guaranteed, provided the PPMs are correct, in the uncalibrated case they depend on the matches computed by RANSAC and on the sucess of a non-linear minimization, so they are data-dependent.
Reference papers:
@Article{FusTruVer99, author = {A. Fusiello and E. Trucco and A. Verri}, title = {A Compact Algorithm for Rectification of Stereo Pairs}, year = {2000}, journal = {Machine Vision and Applications}, volume = {12}, number = {1}, pages = {16-22}, pdf = {http://profs.sci.univr.it/\~fusiello/papers/00120016.pdf} }
@Article{FusIrs10, author = {A. Fusiello and L. Irsara}, title = {Quasi-Euclidean epipolar rectification of uncalibrated images}, journal = {Machine Vision and Applications}, year = 2011, volume = 22, number = 4, pages = {663 - 670}, doi = {10.1007/s00138-010-0270-3}, Webpdf = {http://www.diegm.uniud.it/fusiello/papers/mva10.pdf} }