A new accurate and fast algorithm of sub-pixel image registration

In terms of large field view and multi-channel TDICCD remote sensing images, previous interpolation methods and curve fitting methods are not able to achieve the high accuracy and fast registration, and their noise immunity and robustness is not high. For that, this paper presents a more accurate and faster method, iterative pixel interpolation and surface fitting method. Firstly, the method uses the some overlapping pixels between multi-channel images. Secondly, the proposed technique, which is based on the maximization of the correlation coefficient function, combines an efficient pixel-moving interpolation scheme with surface fitting, which makes use of accurate interpolation calculation and fast surface fitting in the iterative process. Finally, the accuracy and speed of the algorithm is evaluated by sub-pixel registration of multi-channel images and comparison with other sorts of efficient methods. The experiment results show that the accuracy of the method reaches 0.01 pixels and it is 3 times faster than the interpolation method. In the registration of large field view and multi-channel TDICCD images, the method is accurate and fast, with greatly high stability, noise immunity and robustness.