Comparing Images Using SIFT

Question

I am trying to compare 2 images taken from a digital camera. Since there may be movement on the camera, I want to first make the pictures “appropriate” and then compare (using some kind of remote function).

To fit them, I think about cropping the second image and using SIFT to find it inside the first image ... it will probably have a slight difference in scale / translation / rotation so I need to find a transformation matrix that converts image 1 to image 2 (based on points found by SIFT)

any ideas on how to do this (or am I assuming a common problem that some open source implementation might have?)?

thanks

Answer 1

If the motion is small or mostly translational, it may be easier to use phase correlation or just compare the Fourier moments. http://en.wikipedia.org/wiki/Phase_correlation Also check this question. How to use the SIFT algorithm to calculate how similar the two images are?