The Bi-directional Framework for Unifying Parametric Image Alignment Approaches
In this paper, a generic bi-directional framework is proposed for parametric image alignment, that extends the classification of . Four main categories (Forward, Inverse, Dependent and Bi-directional) form the basis of a consistent set of subclasses, onto which state-of-the-art methods have been mapped. New formulations for the ESM  and the Inverse Additive  algorithms are proposed, that show the ability of this framework to unify existing approaches. New explicit equivalence relationships are given for the case of first-order optimization that provide some insights into the choice of an update rule in iterative algorithms.
KeywordsImage Alignment Inverse Additive Reference Coordinate Frame Inverse Compositional Forward Additive
- 2.Benhimane, S., Malis, E.: Real-time image-based tracking of planes using efficient second-order minimization. In: IROS 2004, Sendai, Japan, vol. 1, pp. 943–948 (2004)Google Scholar
- 4.Lucas, B., Kanade, T.: An iterative image registration technique with an application to stereo vision. In: IJCAI 1981, pp. 674–679 (1981)Google Scholar
- 6.Baker, S., Gross, R., Matthews, I., Ishikawa, T.: Lucas-Kanade 20 years on: A unifying framework: Part 2. Technical Report CMU-RI-TR-03-01, Robotics Institute, Carnegie Mellon University, Pittsburgh, PA (February 2003)Google Scholar
- 12.Bartoli, A.: Groupwise geometric and photometric direct image registration. IEEE Transactions on Pattern Analysis and Machine Intelligence (2008) preprint (January 14, 2008)Google Scholar