colmap exhaustive matcher vs spatial matcher

108 views
Skip to first unread message

owen...@gmail.com

unread,
Jul 4, 2020, 2:26:25 PM7/4/20
to COLMAP

I ran two tests with same dataset; about 1200 UAV images. I did the test with the exhaustive matcher; that creates less points in the point cloud 19,488,458 points in this case. As you can see from the screenshot from meshlab, the exhaustive matcher's point cloud are all more closely related; small rectangle in the back of the image.



The spatial matcher on the other hand has way more points 65,765,217; but they are all over the place; as you can see from the much larger rectangle; both models overlap.


snapshot00.png


Is this normal behavior? would it be recommended to just use exhaustive matcher although the FAQ says with GPS data the spatial matcher would be recommended.

Gabriel Hein

unread,
Jul 4, 2020, 4:33:50 PM7/4/20
to col...@googlegroups.com
This seems like normal behavior to me. All pure SfM approaches will return results with arbitrary orientation and scale.

The difference between the spatial matcher and exhaustive matcher is the spatial matcher can use GPS data to only check images nearby to each other for matches. This can save a tremendous amount of processing time over the exhaustive matcher which checks every image against every other image. It can also help avoid incorrect matches. 

I'm not sure about how this is implemented in colmap but the spatial matcher may also use the GPS data to get true scale. But even if it doesn't, there is no reason that two independent reconstructions if there same scene would necessarily have similar scale. One model might be 1m across and one might be 1km - that is expected.

I believe this is also the answer to your previous question about the vertical axis being the reverse of what you want. COLMAP has no idea which way is up so, in general, you'll need to align (and scale) your model after the fact. The spatial matcher with GPS may get the scale correct for you and also get the two horizontal axes in the plane you want but you may still need to flip the vertical axis. Again, I'm not actually sure how COLMAP uses the GPS data - that model may also have arbitrary scale and orientation.





--
You received this message because you are subscribed to the Google Groups "COLMAP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to colmap+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/colmap/a99f188d-972e-4435-a739-0720330de8f5o%40googlegroups.com.
snapshot00.png
Reply all
Reply to author
Forward
0 new messages