Experiments on robust image registration using a markov-gibbs appearance model

Authors:
Ayman El-Baz;Aly Farag;Georgy Gimel’farb
Affiliations:
Computer Vision and Image Processing Laboratory, University of Louisville, Louisville, KY;Computer Vision and Image Processing Laboratory, University of Louisville, Louisville, KY;Department of Computer Science, University of Auckland, Auckland, New Zealand
Venue:
SSPR'06/SPR'06 Proceedings of the 2006 joint IAPR international conference on Structural, Syntactic, and Statistical Pattern Recognition
Year:
2006

Citing 8
Cited 0

Performance Evaluation of Scene Registration and Stereo Matching for Artographic Feature Extraction

IEEE Transactions on Pattern Analysis and Machine Intelligence - Special issue on interpretation of 3-D scenes—part II
Image registration using a new edge-based approach

Computer Vision and Image Understanding
A Knowledge Based Approach to Automatic Image Registration

ICIP '97 Proceedings of the 1997 International Conference on Image Processing (ICIP '97) 3-Volume Set-Volume 3 - Volume 3
Alignment by maximization of mutual information

Alignment by maximization of mutual information
Distinctive Image Features from Scale-Invariant Keypoints

International Journal of Computer Vision
Expectation-Maximization for a Linear Combination of Gaussians

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 3 - Volume 03
Texture Analysis by Accurate Identification of Simple Markovian Models

Cybernetics and Systems Analysis
Extension of phase correlation to subpixel registration

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

A new approach to align an image of a textured object with a given prototype under its monotone photometric and affine geometric transformations is experimentally compared to more conventional registration algorithms. The approach is based on measuring similarity between the image and prototype by Gibbs energy of characteristic pairwise co-occurrences of the equalized image signals. After an initial alignment, the affine transformation maximizing the energy is found by gradient search. Experiments confirm that our approach results in more robust registration than the search for the maximal mutual information or similarity of scale-invariant local features.