Image restoration method based on DU-GAN network

The invention discloses an image restoration method based on a DU-GAN network. The method comprises the steps that 1, mask position coding is added into a generator to accelerate the convergence speed of the network and the accuracy of filling content; 2, adding an attention mechanism into the generator network to enable the texture of the image to be clearer; and 3, the adversarial loss is increased, the content of the generated image is further constrained, and the quality of the restored image is improved. When the damaged image is repaired under the condition that the texture and structure information of the damaged image is relatively complex, the texture and structure information contained in the damaged image can be fully utilized, and the texture and structure information are combined for use, so that the quality of the repaired image is higher, and the visual effect is more natural. 本发明公开了一种基于DU-GAN网络的图像修复方法，包括：1：在生成器中加入掩模位置编码加快网络的收敛速度和填充内容的准确性；2：在生成器网络中加入注意力机制使图像的纹理更加清晰；3：增加对抗损失，进一步约束生成图像的内容，提高修复图像的