Image is the main carrier of information transmission and presentation today, and high-quality image provides a solid foundation for the development of computer vision field because of its clear picture quality and rich details. In this paper, we take deep learning technology as the basis and optimize the image super-resolution reconstruction algorithm with targeted training, adopt enhanced generative adversarial network, and combine the image cross-level self-similarity features to construct a high-quality image super-resolution reconstruction model based on deep convolutional generative adversarial network. Simulation results show that the algorithm proposed in this paper has better performance, the objective evaluation indexes of the reconstructed image are significantly improved, and the operation speed is relatively fast, the recovered image has more high-frequency detail information, and the visual effect is better.
