Semantic Segmentation Aided Image Transmission for Maritime UAV Surveillance Network

In maritime surveillance networks, limited energy consumption resources and long delays are serious issues that deteriorate the performance of image transmission for unmanned aerial vehicles (UAVs). Therefore, in this paper, we provide a task-transfer UAV system based on hybrid networks and a semantic segmentation-aided image transmission scheme. Our work mainly includes three aspects. Firstly, we compare the image reconstruction quality and transmission delay of our proposed technique to those of traditional image transmission schemes. Secondly, the energy consumption of the task-transfer UAV system based on hybrid networks is compared to that of the single-network UAV system. Finally, an energy consumption optimization model is established for the task-transfer UAV system based on hybrid networks under delay and bandwidth constraints. The experimental results show that the proposed image transmission scheme can achieve better anti-noise performance and less latency, and our task-transfer UAV system based on hybrid networks uses less energy on average, which is more suitable for multi-task demand scenarios. In addition, by adjusting the bandwidth of the task-transfer UAV system based on hybrid networks, the energy consumption of the transmission system can be further reduced.