Generation of digital artistic images based on GANs

DOI

10.2991/978-94-6463-266-8_47How to use a DOI?

Keywords

defogging algorithm; Generating a countermeasure network; Image generation; loss function

Abstract

This research proposes an end-to-end defogging algorithm based on Conditional Generative adversarial network (CGAN), aiming at the problem that the image quality of visible light observation equipment is significantly reduced in the case of fog (haze). This algorithm innovatively constructs a new generator that replaces traditional maximum pooling with soft pooling in the convolutional layer of the encoder to improve the ability to extract fine-grained features. At the same time, a global average pooling layer is introduced between the encoder and decoder to eliminate the oscillation effect of image edges, thereby improving the clarity of the defogged image. In addition, the structure of the discriminator based on PatchGAN is simplified, and the design of the Loss function and the selection of the weight value are optimized, thus improving the training speed and quality of the network model. The experimental results show that the algorithm exhibits stable performance in both synthetic and real foggy images. The image after defogging has an improvement in subjective perception of clarity, edge sharpness, color naturalness, and detail restoration. In terms of quantitative indicators, compared to traditional algorithms, this algorithm has improved to varying degrees in terms of structural similarity, peak signal-to-noise ratio, and image information entropy. In summary, the CGAN based end-to-end defogging algorithm proposed in this study has significant effects on improving image quality in foggy (hazy) conditions. By optimizing the network structure, Loss function and weight selection, the algorithm shows good performance in quantitative indicators and target recognition tasks.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Fang Wang
PY  - 2023
DA  - 2023/10/10
TI  - Generation of digital artistic images based on GANs
BT  - Proceedings of the 2nd International Conference on Intelligent Design and Innovative Technology (ICIDIT 2023)
PB  - Atlantis Press
SP  - 438
EP  - 444
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6463-266-8_47
DO  - 10.2991/978-94-6463-266-8_47
ID  - Wang2023
ER  -