MDCN: Multi-Scale Dense Cross Network for Image Super-Resolution

Convolutional neural networks have been proven to be of great benefit for single-image super-resolution (SISR). However, previous works do not make full use multi-scale features and ignore the inter-scale correlation between different upsampling factors, resulting in sub-optimal performance. Instead blindly increasing depth network, we are committed mining image learning factors. To achieve this, propose a Multi-scale Dense Cross Network (MDCN), which achieves performance with fewer parameters less execution time. MDCN consists dense cross blocks (MDCBs), hierarchical feature distillation block (HFDB), dynamic reconstruction (DRB). Among them, MDCB aims detect maximize flow at scales, HFDB focuses on adaptively recalibrate channel-wise responses distillation, DRB attempts reconstruct SR images factors single model. It is worth noting that all these modules can run independently. means selectively plugged into any CNN model improve Extensive experiments show competitive results SISR, especially task multiple The code provided https://github.com/MIVRC/MDCN-PyTorch .