in order to get a homography between two images, I wanted to know how many points I need at least to get a good estimation of the homography matrix?
You need a minimum of four to get anything at all, but I've found 7 to 10 to be a good number in practice, provided that they are chosen "manually". Using some kind of feature detector, you may need ten or 20 times that to find a suitable sub-set with few outliers.
It really depends on the distribution of the points - if you have 10 co-linear points, you're not going to get a happy result, for example. However, for a slightly distorted rectangle mapping, the four corners are all you need.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With