Cloud detection of GF‐7 satellite laser footprint image

Abstract

Abstract In November 2019, the GaoFen‐7(GF‐7) satellite was equipped with China's first laser altimeter with full waveform recording capability, which obtains high‐precision long‐range three‐dimensional coordinates. The influence of clouds is noticeable for laser transmission, and a footprint camera is used to determine laser pointing and to image the ground. However, the cloud inevitably appears in the laser footprint image. In this study, the authors propose a cloud detection scheme for footprint images based on deep learning. First, an adaptive pooling model is proposed according to the characteristics of the cloud region. Next, model fusion was performed based on the SegNet and U‐Net training results. Finally, test time augmentation was used to enhance the data and to improve cloud detection accuracy. The experimental results show that the fusion result of the model was approximately 5% better than that of the traditional cloud detection algorithm, which improved the shortcomings of the traditional algorithm, such as poor detection effect for thin clouds and complex underlying cloud surfaces. The related conclusions have certain reference significance for GF‐7 data processing and related research on footprint images.

Keywords

• Cloud physics
• Data and information; acquisition, processing, storage and dissemination in geophysics
• Instrumentation and techniques for geophysical, hydrospheric and lower atmosphere research
• Optical, image and video signal processing
• Atmospheric, ionospheric and magnetospheric techniques and equipment
• Computer vision and image processing techniques