SSM-CloudNet: state space model-based cloud and cloud shadow detection network for Sentinel-2 imagery

Abstract Remote sensing images acquired by satellite sensors are inevitably obscured by clouds and cloud shadows, which obscure true ground reflectance information in those areas. Therefore, to ensure the accuracy and reliability of downstream analyses, detecting clouds and their shadows in satellite images is crucial. This paper proposes a state space model-based cloud and cloud shadow detection network for Sentinel-2 imagery, named SSM-CloudNet. This is the first time that a state space model (SSM) is applied to cloud and cloud shadow detection tasks, combining the efficient feature extraction capability of the symmetric encoder-decoder structure with the accuracy of SSM in cloud feature extraction, and providing an innovative approach to this task. Experiments demonstrate that SSM-CloudNet performs exceptionally well across multiple metrics, proving the high effectiveness of the proposed method.