Abstract
Nowadays, the need for high-quality image reconstruction and restoration ismore and more urgent. However, most image transmission systems may suffer fromimage quality degradation or transmission interruption in the face ofinterference such as channel noise and link fading. To solve this problem, arelay communication network for semantic image transmission based on sharedfeature extraction and hyperprior entropy compression (HEC) is proposed, wherethe shared feature extraction technology based on Pearson correlation isproposed to eliminate partial shared feature of extracted semantic latentfeature. In addition, the HEC technology is used to resist the effect ofchannel noise and link fading and carried out respectively at the source nodeand the relay node. Experimental results demonstrate that compared with otherrecent research methods, the proposed system has lower transmission overheadand higher semantic image transmission performance. Particularly, under thesame conditions, the multi-scale structural similarity (MS-SSIM) of this systemis superior to the comparison method by approximately 0.2.