A Subspace Method for Fourier-Based Image Registration

Image registration is the process of estimating the misalignment between two images. Automatic image registration procedures are important in many remote-sensing applications. A popular approach for coarse registration is Fourier based. Fourier-based image registration is essentially a frequency estimation problem in the frequency domain of the image data. This letter studies the application of a subspace-based frequency estimation approach for the Fourier-based image registration problem. By employing the Multiple Signal Classifier algorithm, a more robust and accurate registration result is achieved at the expense of moderate computational complexity. In the examples considered, performance for registration of remote-sensing images using this method is comparable to a mutual information-based and a correlation-based registration algorithm but requires less computation.