Hybrid Recursive Super-resolution Image Reconstruction using Neural Networks

Most published super-resolution image reconstruction techniques usually perform well on small magnifications, however, as magnification increases, they suffer two severe problems: the exponentially increased computational complexity and the degradation of reconstruction quality caused by severe ringing artifacts. In this paper, a hybrid recursive reconstruction algorithm using neural networks is proposed to tackle these two problems.