The image registration of the global motion model is used successfully to the image super-resolution reconstruction. It can estimate motion parameters more accurately than the optimized registration algorithm in the frequency domain if the motion model is a good approximation to the motion between images

therefore

can insure that the reconstructed high resolution image possesses more details. At the same time

it is expounded that the larger the frame number of the low resolution images is

the more accurate the reconstructed image will be. But with the increasement of the frame number of the LR images

the diversification of the mean square error between the reconstructed image and the real image diminishes

and cost of the algorithm would be increased greatly. So

the frame number of the LR images should be determined according to the precision of the reconstructed image.