Super-Resolution Algorithm Based on Transformer+CNN

CNN and Transformer have their own excellent performance in image super-resolution, but these methods are difficult to be applied to the field of image SR alone due to the challenges of balancing model performance and complexity. To address these issues, we propose a Transformer+CNN-based algorithmic model for image super-resolution. First, a channel attention module is added to the model framework to model the correlation between channels of the feature maps; second, different training strategies are used to fine-tune the parameters at the end of the network base training; finally, a new loss function is introduced to optimize the parameters in the fine-tuning phase. The experimental results show that the method is more effective than the excellent algorithms in recent years.