can someone please elaborate a bit more on how we know the minimum number of sample points to fit the model?
motoole2
The minimum number of sample points will vary based on the problem.
When fitting lines to a bunch of random points (see this slide), we needed two points to fit our line model. And as a result, the number of samples in this case is two.
When computing translation between pairs of frames (as in this slide), we again need two points, representing one correspondence. This is because a minimum of two points is required to define a translation operation.
Finally, for homographies (as in this slide), we require a minimum of four points. This is, once again, because a minimum of four points is required to compute a homography uniquely.
can someone please elaborate a bit more on how we know the minimum number of sample points to fit the model?
The minimum number of sample points will vary based on the problem.
When fitting lines to a bunch of random points (see this slide), we needed two points to fit our line model. And as a result, the number of samples in this case is two.
When computing translation between pairs of frames (as in this slide), we again need two points, representing one correspondence. This is because a minimum of two points is required to define a translation operation.
Finally, for homographies (as in this slide), we require a minimum of four points. This is, once again, because a minimum of four points is required to compute a homography uniquely.