Knowledge Distillation Based Fast Degradation Estimator For Blind Super-resolution
Lei Jian, Ying Chun Zhong, Xiao Feng Du · 2023
Blind Super Resolution aims to enhance low-resolution images to high-resolution ones without the need to initially understand the detailed information of the original low-resolution images. In recent years, Blind Super Resolution techniques have attempted to guide image super-resolution reconstruction by designing explicit degradation estimators. However, these degradation estimators often come with a large number of parameters, slow processing speed, and less-than-ideal out-comes. In this study, we introduce an improved fast degradation estimator based on knowledge distillation, enabling the rapid and effective extraction of unknown degradation information from images. This estimator further guides the Super- Resolution(SR) network in achieving image super-resolution reconstruction. Ex-perimental results indicate that our enhanced fast degradation estimator doubles its operational efficiency and achieves super-resolution reconstruction for low-resolution images affected by various degradation factors. This advancement holds the promise of delivering superior performance and results for image super-resolution reconstruction tasks.