Research on Artificial Intelligence Art Image Synthesis Algorithm Based on Generation Model

As an important artistic creation form and national heritage, painting shows the features of human culture and conveys human thoughts and feelings through beautiful forms of expression, which is a unique form of human understanding of the world. Image generation algorithm is actually the transformation from one image to another, and image synthesis is an important research direction in the field of computer vision. In this paper, the AI (artificial intelligence) art image synthesis algorithm based on generation model is studied. By studying the development of GAN (Generative Adversary Networks), this paper proposes an AI art image synthesis algorithm based on GAN. An hourglass-shaped CNN (Convolutional Neural Network) with layer hopping connection is designed, which consists of coding and decoding. By adding corresponding interval convolution layers, the underlying texture information of the image is preserved, and the identity loss is maintained, so that the original image and the generated image are consistent in color. The results show that all three CNN models have high recognition accuracy for data sets, and the recognition accuracy for both data sets is higher than 99%. The algorithm in this paper is superior to other comparison algorithms in RMSE and PSNR. Experimental results show that this method improves the quality of generated images.