通信与信息技术Issue(3)：51-56,6.
计算机视觉驱动的视频通信图像超分辨率重建技术研究
Research on computer vision-driven super-resolution reconstruction technology for video communication images
摘要
Abstract
With the rapid development of scenarios such as 5G communication,remote office and intelligent security,the demand for image resolution in video communication continues to increase.However,issues like low bandwidth constraints,complex motion interfer-ence and multi-type noise often lead to quality defects in the transmitted images,such as detail loss and blurriness.This paper takes com-puter vision technology as the core driver to study the technical architecture,bottleneck breakthrough paths and application implementa-tion methods of super-resolution reconstruction for video communication images.By constructing a scene-aware dynamic feature weight allocation mechanism,a lightweight Transformer feature fusion architecture and a multi-scale noise adaptive suppression module,the bal-ance between reconstruction accuracy and real-time performance is achieved.关键词
计算机视觉/动态特征适配/帧间一致性维护/多噪声抑制Key words
Computer vision/Dynamic feature adaptation/Inter-frame consistency maintenance/Multi-noise suppression分类
信息技术与安全科学引用本文复制引用
胡淇程,关长辉,任鑫宇,金岩,雷强,郑伟南..计算机视觉驱动的视频通信图像超分辨率重建技术研究[J].通信与信息技术,2026,(3):51-56,6.基金项目
吉林省高校创新训练项目:"全国首个提出基于YOLO11和卡尔曼滤波目标检测的河道巡检无人机视觉检测系统"(项目编号:S202513604061) （项目编号:S202513604061）
