PDF version of this entire document
For automated correspondence across images I have looked at http://people.cs.ubc.ca/ lowe/keypoints/David Lowe's SIFT (for 2-D) and a Bookstein-inspired approach to finding correspondence in 3-D, which was implemented and http://www.mathworks.co.uk/matlabcentral/fileexchange/34066-point-correspondence-among-3d-shapepublished as code one month ago. A quick look at the code reveals a fairly low quality and there is also a requirement that normals for each point get supplied.
There is an interesting recent paper on finding 3-D corresponded at
http://www.waset.org/journals/ijeee/v3/v3-3-21.pdfhttp://www.waset.org/journals/ijeee/v3/v3-3-21.pdf
Kaiser et al. combined this with texture [16].
