Pre-matching: the problem!

If the sensors where square I wouldn’t have a problem but now they aren’t.

Let’s say one image covers 25 meters on the hight and 35 at the width, that means that if you set the front and side overlap to the same percentage it wouldn’t match in overlap per meter.

It also means that setting the prematch distance and neighbours is a bit messy. If you set the distance with the image hight you would loose some overlapping images on the side, if you would set it from the width you would get false positives, images that doesn’t overlap in the front. So if you set from the hight you loose some overlapping images but reduce processing time and from the width you gain matches and increase processing time.

If I loose a few matching images, what do I loose in the end result?

You’re not not matching them, it’s adjusting the bundle size and pool of what gets matched against each other.

Every image should be matched to something else by the end of the stage.

1 Like

But it might take longer to process

Every image needs to be matched with at least one other image for the common features to be used for reconstruction, though.

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.