一种基于深度学习的图像风格化算法

In the field of image processing, image stylization is a kind of technology that processing image information such as the colour, silhouette, line to change image effect by computer.In recent years, with the development of machine learning, image stylizationprocessing based on neural network obtained good effect. Based on the deep neural network, on the basis of researching the existing network and integrating the concept of generative adversarial networks, this paper designs a network to generate a highly stylized effect pictures. A scoring test model was also trained to test whether the effect of the generated image was similar to the real type of image,and to compare the generated images of two networks objectively. The experimental results show that the new network has better effect and can generate more realistic images.