1 | initial version |

RANSAC gets as input a set of matches and randomly selects some of them to compute a homography H. It then tests, how many matches 'fit' to this homography. If a match is good enough, it counts as an inlier. This is repeated many times and hopefully, RANSAC will find a homography that has many inliers and thus a good score.

The above formula is the test that defines, whether a match is an inlier or not.
Given a match `m(srcPoint, dstPoint)`

and a homography `H`

, m is an outlier, if the distance between `dstPoint`

and `H*srcPoint`

(homography applied to `srcPoint`

) is greater than `ransacReprojThreshold`

.

Copyright OpenCV foundation, 2012-2018. Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.