Research Article
Colorization of Characters Based on the Generative Adversarial Network
@INPROCEEDINGS{10.1007/978-3-030-48513-9_51, author={Changtong Liu and Lin Cao and Kangning Du}, title={Colorization of Characters Based on the Generative Adversarial Network}, proceedings={Cloud Computing, Smart Grid and Innovative Frontiers in Telecommunications. 9th EAI International Conference, CloudComp 2019, and 4th EAI International Conference, SmartGIFT 2019, Beijing, China, December 4-5, 2019, and December 21-22, 2019}, proceedings_a={CLOUDCOMP}, year={2020}, month={6}, keywords={Image colorization Intelligent electric grid Generative Adversarial Network}, doi={10.1007/978-3-030-48513-9_51} }- Changtong Liu
Lin Cao
Kangning Du
Year: 2020
Colorization of Characters Based on the Generative Adversarial Network
CLOUDCOMP
Springer
DOI: 10.1007/978-3-030-48513-9_51
Abstract
With the development of economy, global demand for electricity is increasing, and the requirements for the stability of the power grid are correspondingly improved. The intelligence of the power grid is an inevitable choice for the research and development of power systems. Aiming at the security of the smart grid operating environment, this paper proposes a gray-scale image coloring method based on generating anti-network, which is used for intelligent monitoring of network equipment at night, and realizes efficient monitoring of people and environment in different scenarios. Based on the original Generative Adversarial Network, the method uses the Residual Net improved network to improve the integrity of the generated image information, and adds the least squares loss to the generative network to narrow the distance between the sample and the decision boundary. Through the comparison experiments in the self-built CASIA-Plus-Colors high-quality character dataset, it is verified that the proposed method has better performance in colorization of different background images.