search

paolodonadei / sfmh1

Automatically exported from code.google.com/p/sfmh1
0 stars 0 forks source link

bad matches remaining #12

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago 
What steps will reproduce the problem?
1. use sift not oxford matches
2. reconstruct
3. you will see points int he middle of nowhere, points that are clearly
outliers

i should go through their error with respect to the fundamental matrix and
figure out why they are not getting filtered then do something about them,
they are messing up the reconstruction and also the addition of new frames,
if these bad matches remain they will mess up everything

Original issue reported on code.google.com by houman.r...@gmail.com on 18 May 2010 at 10:24