Research Article
Simulation Generation Algorithm for Foggy Images in Natural Scenes
@INPROCEEDINGS{10.1007/978-3-031-31733-0_16, author={Jianping Liu and Qing Ye and Shizhuo Qiu and Yuze Liu}, title={Simulation Generation Algorithm for Foggy Images in Natural Scenes}, proceedings={Smart Grid and Innovative Frontiers in Telecommunications. 7th EAI International Conference, SmartGIFT 2022, Changsha, China, December 10-12, 2022, Proceedings}, proceedings_a={SMARTGIFT}, year={2023}, month={5}, keywords={Fog Simulation Depth Estimation Automatic Inspection}, doi={10.1007/978-3-031-31733-0_16} }- Jianping Liu
Qing Ye
Shizhuo Qiu
Yuze Liu
Year: 2023
Simulation Generation Algorithm for Foggy Images in Natural Scenes
SMARTGIFT
Springer
DOI: 10.1007/978-3-031-31733-0_16
Abstract
The foggy environment seriously affects the automatic inspection or cruise monitoring of outdoor equipment in the power system. Aiming at the lack of fog image data sets, a fog simulation image generation algorithm based on depth estimation was proposed. First of all, by unsupervised depth estimation model building the depth map of an outdoor clear picture. Then using feature fusion to refine the depth chart details. By setting the atmospheric extinction coefficient for transmittance figure. Dark channel method is used to estimate the atmospheric light value of the image. Finally, the fog simulation images are obtained based on the atmospheric scattering model. The experimental results show that the method improves the depth map effectively, and the generated fog simulation image is reliable. The average error rate of fog simulation is 6.2%, which solves the problem of excessive uneven fog edge, and the fog simulation image effect is very good in low visibility. Generating fog images with different visibility labels can solve the problem of lack of fog datasets.
