realhire's profile - activity

I have already learned a lot about how to do 3D reconstrction from 2 views. My question is very similar to this one: http://stackoverflow.com/questions/36...

However, in this post, not any details are given. In this post: "Suppose I have 20 images around a static, rigid body all taken with the same camera." The answer suggests:"You start by obtaining pairwise calibration P1-P2, P2-P3, P3-P4 ... using feature points of the corresponding pair. " I can do that, but I don't know how to combine these feature points into the same the common coordinate frame. The connecting between P1-P2, P2-P3 is not a trival problem, but not many textbooks are talking about it. In the book"Mastering OpenCV with Practical Computer Vision Projects", it said something about "Perspective-N-Point", but I cannot understand it. In this method, it seems that they use 3D-2D pairs to computer the P, but I got the local P. If I have known the local P, why I spend the time to find 3D-2D pair to get the P again.

This is my question: I want to project the 2D pair points to the common coordinate frame by the local P and the related Pi.

I don't know the function of solvePnP???????

If you can give me some examples or equations, this will great!