NMS-based blurred image sub-pixel registration

Sub-pixel accurate registration is the challenging aspect of super-resolution reconstruction. However, the blur contained in the low-resolution images will result in inaccurate registration and an unsatisfactory reconstruction result. Therefore, this paper introduces a blurred image sub-pixel registration method based on Non-Maxima Suppression (NMS). By analyzing the relationship between the image edge energy and the registration precision, NMS technology is used to extract the local gradient magnitude maximum point set adaptively, which is used for image registration as strong information. Experimental results show that this method can improve the registration accuracy of blur low-resolution images.