Two-stage correlation method in low SNR for fast image registration

Abstract

A two-stage correlation method which can realize fast registration of two digital images in the case of low SNR (signal-to-noise ratio) is proposed, and a new algorithm for roughly searching for a registration position-sequential measure test algorithm (SMTA) is given. Monte-Carlo simulation has shown that the method is over two orders of magnitude faster than the optimal MSD algorithm and far faster than the sequential SSDA, and that its performance is almost as good as the optimal MSD. It can perform well under the condition that the SNR is so low and the length of subimage data is so short that SSDA cannot be used to match at all. The fast correlation method virtually satisfies the demands of some navigation, tracking, and guidance systems for real-time matching.<>