An enhanced image registration algorithm by the Burg filter

The problem of registration of displaced images is considered. A well known method for this problem is the phase-correlation algorithm, which provides superior performance when rigid plane translation may be expected, even if the images are noisy and subject to varying conditions of illumination. However, the computational cost of this approach is high since it requires complex arithmetics. The computational cost can be reduced by using the correlation function between the binary boundary maps of the images, computed by Fermat number transforms. Based on this approach, an enhanced method which uses the Burg filter to whiten the image sequences before the contour extraction is proposed. The experimental results prove the superior performance of the proposed method with respect to direct correlation.<
>