Satellite-ground communication is a critical component in the global communication system, significantly contributing to environmental monitoring, radio and television broadcasting, aerospace operations, and other domains. However, the technology encounters challenges in data transmission efficiency, due to the drastic alterations in the communication channel caused by the rapid movement of satellites. In comparison to traditional transmission methods, semantic communication (SemCom) technology enhances transmission efficiency by comprehending and leveraging the intrinsic meaning of information, making it ideal for image transmission in satellite communications. Nevertheless, current SemCom methods still struggle to adapt to varying channel conditions. To address this, we propose a SemCom transmission model based on a Channel Code-Book (CCB) for adaptive image transmission in diverse channel environments. Our model reconstructs and restores the original image by documenting fading and noise states under various channel conditions and dynamically adjusting the denoiser's model parameters. Extensive experimental results demonstrate that our CCB model outperforms three representative baseline models, including Deep JSCC, ASCN, and WITT in various environments and task conditions, achieving an advantage of more than 10 dB under high signal-to-noise ratio conditions.
Keywords: channel adaptive; image transmission; satellite–ground scenarios; semantic communication.