| Summary: | In this age of rapid development of science and technology, people get information in various ways, from computers, from mobile phones, from books. Among them, image as an important carrier of information transmission. Its importance in people's life is self-evident. Image distortion caused by image folding is a new research field. After being repaired by the Markov random field method, the image still has
distortion. This dissertation mainly studies deep learning methods and traditional
image processing methods to further solve image distortion, which involves relevant theoretical knowledge in the field of image super resolution, including convolutional neural networks, residual networks, dense networks and various classical image super resolution networks. In this dissertation, the VDSR model is modified and a good result is obtained.
|
|---|