摘要
This paper presents a closed-form robust phase correlation based algorithm for performing image registration to subpixel accuracy.The subpixel translational shift information is directly obtained from the phase of the normalized cross power spectrum by using Maximum Likelihood Estimation(MLE).The proposed algorithm also has slighter time complexity.Experimental results show that the proposed algorithm yields superior registration precision on the Cramér-Rao Bound(CRB) in the presence of aliasing and noise.
This paper presents a closed-form robust phase correlation based algorithm for performing image registration to subpixel accuracy. The subpixel translational shift information is directly ob- tained from the phase of the normalized cross power spectrum by using Maximum Likelihood Esti- mation （MLE）. The proposed algorithm also has slighter time complexity. Experimental results show that the proposed algorithm yields superior registration precision on the Cram@r-Rao Bound （CRB） in the presence of aliasing and noise.
