Image dehazing algorithm based on ViTGAN and its application

Abstract In recent years, many methods based on deep learning have emerged in the field of image dehazing. The ViTGAN generation adversarial network Transformer is used to design dehazing algorithm.In ViTGAN, generator improved achieve desired effect.The first half collaterals can be as a data encoder extract features original image. this part, lower sampling layer placed behind each density block.Generate regard second corresponding decoder, block a, for passed characteristics sampling, reduced by enlarge size, ensure that output correct.The feature map decoder and symmetric fused properly represented.