面向计算机视觉任务的无线图像传输  

Wireless Image Transmission for Computer Vision Task

作　　者：王丽娟 吴晓红[1] 杨红[1] 卿粼波[1] WANG Lijuan;WU Xiaohong;YANG Hong;QING Linbo(College of Electronics and Information Engineering,Sichuan University,Chengdu Sichuan 610065,China)

机构地区：[1]四川大学电子信息学院,四川成都610065

出　　处：《通信技术》2025年第1期19-25,共7页Communications Technology

基　　金：四川省自然科学基金青年基金(2024NSFSC1419)。

摘　　要：随着信息技术的融合创新及智慧城市等领域的迅猛发展,机器视觉任务和图像数据传输需求逐渐增加。语义通信技术通过提取和传输数据的语义信息,而非原始比特流,以提高数据的传输效率。卷积神经网络(Convolutional Neural Networks,CNN)和Transformer结构的结合为图像语义通信领域带来了新的突破。CNN在提取图像局部特征方面表现出色,而Transformer则擅长捕捉长距离依赖和全局特征。聚焦于图像的传输与处理,基于CNN和Transformer结构对于特征提取的优势,并加入注意力机制,提出了一种高效的端到端图像语义通信方案,以在优化传输性能的同时满足智能任务对图像数据传输的需求。结果显示,所提方案与现有方法相比不仅提高了计算效率,还具有更强的鲁棒性和适应性。With the integration and innovation of information technology and the rapid development of smart cities and other fields,the requirements of machine vision tasks and image data transmission is gradually increasing.Semantic communication technology improves data transmission efficiency by extracting and transmitting semantic information of the data instead of the original bit stream.The combination of CNN(Convolutional Neural Network) and Transformer structure brings new breakthroughs in the field of image semantic communication.CNN excels at extracting local features from images,while Transformer excels at capturing long-distance dependencies and global features.This paper focuses on the transmission and processing of images,based on the advantages of CNN and Transformer structures for feature extraction and incorporating the attention mechanism,it proposes an efficient end-to-end image semantic communication scheme to meet the requirements of intelligent tasks for image data transmission while optimizing the transmission performance.Experimental results indicate that the proposed scheme not only improves the computational efficiency but also has stronger robustness and adaptability compared with the existing methods.

关 键 词：计算机视觉 语义通信 深度学习 无线图像传输 

分 类 号：TN929.5[电子电信—通信与信息系统]