SUPER RESOLUTION IMAGE RECONSTRUCTION FROM LOW RESOLUTION ALIASED IMAGES

Super resolution(SR) algorithms reconstruct a higher resolution image from a set of low resolution images of a scene. Precise alignment of image set is the essential part of these algorithms. If the low resolution images are undersampled and have aliasing effects, performance of the registration algorithms will decrease. In this paper, a new frequency domain approach to precisely registration of low resolution images is presented and it is extended to the registration of aliased images. In this method, an aliasing-free part in the low frequency band is used for registration of these aliased images. A high resolution image is then reconstructed by using bicubic interpolation. This technique is analysed on the test images and results have been discussed.