Decoupling Induction and Multi-Order Attention Drop-Out Gating Based Joint Motion Deblurring and Image Super-Resolution

Resolution decrease and motion blur are two typical image degradation processes that usually addressed by deep networks, specifically convolutional neural networks (CNNs). However, since real images obtained through multiple degradations, the vast majority of current CNN methods employ a single process inevitably need to be improved account for effects. In this work, motivated decoupling multiple-order attention drop-out gating, we propose joint recovery model efficiently address resolution reduction simultaneously. Our style improves continence efficiency construction training. Moreover, proposed multi-order mechanism comprehensively hierarchically extracts features fuses them properly gating. The approach is evaluated using diverse benchmark datasets including natural synthetic images. experimental results show our method can complete super-resolution (SR).