Image Restoration of Agricultural History Based on Neural Patch Synthesis

In recent years, the study of agricultural history has been constantly promoted. In the process of developing agricultural modernization, more and more attention has been paid to the origin of agriculture. With the development of digital image restoration technology and neural network, this study proposes to repair agricultural propaganda paintings of new China by means of neural patch synthesis. First, use the structure of the encoder and decoder to repair the overall structure of the image, and use the idea of the discriminator to make the overall image realistic. Secondly, the texture information of the globally restored image is further optimized by using the feature layer of VGG network, resulting in better results of both global and local restoration. Finally, the actual discriminator is used for realistic processing. The experimental results show that the restored images produced by this method are lifelike, without obvious artificial boundary, and the restored parts are not blurred, which is suitable for the restoration of agricultural history images.