基于GAN网络的遥感图像去云研究

Research on the Cloud Removal of Remote Sensing Images Based on the GAN Network

主要研究了遥感图像应用中的重要问题,特别是云遮挡可能导致数据准确性下降的情况。为了克服这一问题,采用了组卷积、多头自注意力机制和生成对抗网络,构建了一种端到端的遥感图像去云网络。通过在RICE数据集上的实验证明,研究的方法在解决云层遮挡问题方面表现显著,成功生成了清晰且无云的遥感图像。总体而言,为水利工程和遥感图像处理领域提供了一种高效的解决方案,尤其在空间信息利用、计算和参数优化方面具备明显优势。

This article mainly studies a crucial issue in the application of remote sensing images,especially the situa-tion that cloud occlusion can lead to the degradation of data accuracy.In order to address this problem,this article uses group convolution,the multi-head self-attention mechanism and the generative adversarial network to con-struct an end-to-end cloud removal network of remote sensing images.Experiments on the RICE dataset demon-strate that the proposed approach has a significant effect in solving the problem of cloud occlusion,and successfully generates clear and cloud-free remote sensing images,which provides an effective solution for the field of hydraulic engineering and remote sensing image processing overall,and especially has its obvious advantages in the utilization,computation and parameter optimization of spatial information.