4 March 2015 An efficient framework for modeling clouds from Landsat8 images
Author Affiliations +
Proceedings Volume 9443, Sixth International Conference on Graphic and Image Processing (ICGIP 2014); 94431X (2015) https://doi.org/10.1117/12.2179444
Event: Sixth International Conference on Graphic and Image Processing (ICGIP 2014), 2014, Beijing, China
Abstract
Cloud plays an important role in creating realistic outdoor scenes for video game and flight simulation applications. Classic methods have been proposed for cumulus cloud modeling. However, these methods are not flexible for modeling large cloud scenes with hundreds of clouds in that the user must repeatedly model each cloud and adjust its various properties. This paper presents a meteorologically based method to reconstruct cumulus clouds from high resolution Landsat8 satellite images. From these input satellite images, the clouds are first segmented from the background. Then, the cloud top surface is estimated from the temperature of the infrared image. After that, under a mild assumption of flat base for cumulus cloud, the base height of each cloud is computed by averaging the top height for pixels on the cloud edge. Then, the extinction is generated from the visible image. Finally, we enrich the initial shapes of clouds using a fractal method and represent the recovered clouds as a particle system. The experimental results demonstrate our method can yield realistic cloud scenes resembling those in the satellite images.
© (2015) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Chunqiang Yuan, Chunqiang Yuan, Jing Guo, Jing Guo, } "An efficient framework for modeling clouds from Landsat8 images", Proc. SPIE 9443, Sixth International Conference on Graphic and Image Processing (ICGIP 2014), 94431X (4 March 2015); doi: 10.1117/12.2179444; https://doi.org/10.1117/12.2179444
PROCEEDINGS
5 PAGES


SHARE
Back to Top