Performance Bounds on Image Registration

Registration is a fundamental step in image processing systems where there is a need to match two or more images. Applications include motion detection, target recognition, video processing, and medical imaging. Although a vast number of publications have appeared on image registration, performance analysis is usually performed visually, and little attention has been given to statistical performance bounds. We tackle this problem by deriving theoretical performance bounds for the image registration problem.

Setup: The image shows the landmarks used for registration based on landmarks.

Performance Analysis: The evolution of the variance of the estimated distortion parameters and their Cramer-Rao bounds.
These findings on the variance and CRB can be used to select e.g. the number of points to be used in landmark based registration.

  • We have also derived performance bounds for registration based on intensity values (rather than landmarks) and for more general non-linear distortions (rather than rigid and affine deformations). These results can be seen in the references.

  • References:
    1. I. S. Yetik, "A Novel Non-Rigid Registration Method", to appear, Proceedings of IEEE Southwest Symposium on Image Analysis and Interpretation, 2010.
    2. I. S. Yetik and A. Nehorai, "Performance bounds for image registration," IEEE Transactions on Signal Processing, vol. 54, pp. 1737-1749, 2006.
    3. I. S. Yetik and A. Nehorai, "Performance bounds on image registration," Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing, vol. II, pp. 117-120, 2005.