This paper proposes a novel technique called Keren-FFTGAN for super-resolution image reconstruction from multiple low-resolution images. The technique combines Keren and Fast Fourier Transform (FFT) based image registration methods with a generative adversarial network (GAN) for image reconstruction. The image registration stage uses the Keren algorithm to estimate rotation and translation parameters and the FFT method to estimate scaling parameters between input images. The registered images are then fed into a GAN model consisting of a generator and discriminator network for reconstructing the high-resolution output image. The generator uses an efficient sub-pixel convolution layer to upscale images. Extensive experiments demonstrate that the proposed Keren-FFTGAN technique outperforms state-of-the-art methods in terms of both objective image quality metrics and perceptual quality. The technique proves highly effective in recovering finer details from low-resolution images. By integrating robust image registration and advanced deep learning models, Keren-FFTGAN provides a novel way to boost spatial resolution for various applications.