Translate camera calibration routines to python

[ezwelty]

Translate camera calibration routines to python

[ezwelty]

@douglas-brinkerhoff Finally, I think I've come up with a scheme for camera optimization that is both very flexible and not hideously messy. The meager ransac.py has been replaced by the beefyoptimize.py. I also dropped the image.Camera.optimize() method, since that functionality is now entirely within optimize.py.

In optimize.py, there are the three "camera control" classes:

• Points: image-world point correspondences.
• Lines: image and world lines believed to overlap.
• Matches: image-image point correspondences.

Each of these classes reference image.Camera objects, store the observed image coordinates, and return predicted image coordinates from the current state of the cameras.

The multi-camera model class optimize.Cameras accepts a list of image.Camera, a list of control objects, and the parameters to optimize for each camera separately and for all cameras together. In specific cases (1-2 Cameras, 1 Points or Matches control), it can also be used with optimize.ransac().