Tuesday, 19 March 2013

New Ideas

Its been days since I am trying to find an optimal algorithm to find the transformation between two point clouds with outliers and I have failed so far. I have a new idea.

  • Compute the centroid of the two point clouds (least means squares)
  • Find the displacement of the centroid
  • Repeat these with "hypothetical inliers" and revise the error (RANSAC)
This would give me a pretty good approximation....

No comments:

Post a Comment