A projection-based extension to phase correlation image alignment

Phase-correlation (PC) is a computationally efficient scheme for estimating 2-D and 3-D translations. We present a masking operator that significantly improves the accuracy and robustness of the PC scheme. The masking projects the estimated correlation function on the space of correlation functions that result from a certain range of translations, while rejecting components that are unrelated to the estimated motion. Thus, the registration accuracy is improved by an order of magnitude, especially when registering noisy images and volumes. The scheme is also shown to improve the registration of rotated images in the Fourier domain.