Multiresolution image registration algorithm in wavelet transform domain

Abstract

This paper presents a registration algorithm with a new approach for automatic control point selection within a wavelet transform of astronomical images. It simply looks at the evolution of the coefficients across the different scales and extracts more significant points according to its contribution weight to the multiresolution local contrast entropy. In the proposed algorithm mutual information is applied as a similarity measure to get the best matching between extracted control points. A wavelet multiresolution approach is used to accelerate the matching process while maintaining the accuracy of mutual information. The proposed algorithm has been applied on single sensor astronomical data. The result shows that the algorithm improves computational efficiency and yields robust and consistent image registration results.