# Epipolar geometry estimated using all matches

As you have seen in the previous pages, the epipolar geometry estimated with points from the calibration apparatus is not valid for the scene, and *vice versa*. By combining the two sets of matches (available HERE for the first image and THERE for the second image), we can obtain a new estimation of the epipolar geometry (the fundamental matrix is available HERE).

Now you can play with these two images. You will find
that this newly estimated epipolar geometry is very good for the
scene. The epipolar lines are almost vertical, as we
expected.

What's important is that the epipolar geometry estimated from
all matches is also *valid* for the calibration
apparatus.

Author: **Zhengyou Zhang**

Email: *zzhang@sophia.inria.fr*

Last modified: Wed Oct 23 13:44:28 MET DST