On Jan 9, 4:04 pm, Matt <matt271829-n...@yahoo.co.uk> wrote:
> This is a *horrible* and inefficient method requiring nine sets of > measurements which I'm posting only because there have been no other > suggestions. It appears to work with simulated data but I don't know > how stable it would be in a real-world application.
Ya, getting nine measurements won't be a problem. We typically have hundreds or even thousands of frames of information. However, I also worry about how this method will stand up to noise. Our feature locations can be noisy, so I don't expect an exact solution will come out the way we want it to. But, thanks for the idea.