High-Magnification Super-Resolution Reconstruction of Image with Multi-Task Learning

Single-image super-resolution technology has made great progress with the development of convolutional neural network, but most current methods do not attempt high-magnification image reconstruction; only reconstruction ×2, ×3, ×4 magnification is carried out for low-magnification down-sampled images without serious degradation. Based on this, this paper proposed a single-image method, which extends scale factor to high magnification. By introducing idea multi-task learning, process decomposed into different tasks. Different tasks are trained data, and network models can be obtained. Through cascade task models, low-resolution accumulates advantages layer by layer, we obtain final results. The method shows better performance in quantitative qualitative comparison benchmark dataset than other methods.