电信科学 ›› 2023, Vol. 39 ›› Issue (7): 156-165.doi: 10.11959/j.issn.1000-0801.2023150

• 工程与应用 • 上一篇    

超分辨率重建技术及其在智能终端上的应用

祝谷乔1, 姜超2, 徐煜烨2   

  1. 1 天翼数字生活科技有限公司,上海 200085
    2 中国信息通信科技集团有限公司,湖北 武汉 430070
  • 修回日期:2023-07-14 出版日期:2023-07-20 发布日期:2023-07-01
  • 作者简介:祝谷乔(1976- ),女,天翼数字生活科技有限公司高级工程师,主要研究方向为IPTV关键技术研究与标准化
    姜超(1985- ),男,中国信息通信科技集团有限公司正高级工程师,主要研究方向为多媒体通信、光纤通信与智能终端等
    徐煜烨(1992- ),女,中国信息通信科技集团有限公司工程师,主要研究方向为流媒体传输协议、音/视频编解码等

Super-resolution reconstruction technology and its application on intelligent terminal device

Guqiao ZHU1, Chao JIANG2, Yuye XU2   

  1. 1 E-surfing Digital Life Technology Co., Ltd., Shanghai 200085, China
    2 China Information Communication Technologies Group Co., Ltd., Wuhan 430070, China
  • Revised:2023-07-14 Online:2023-07-20 Published:2023-07-01

摘要:

简单介绍了超分辨率重建技术的发展历程和几种有代表性的超分方法及其实现原理,提出了在智能终端上实现超分辨率重建技术的方案,通过实验仿真,给出了采用插值算法和深度学习算法实现的单图像超分辨率对终端处理性能、图像质量等方面的评估与分析,提出了利用智能终端实现超分的适用场景的建议。并进一步探讨了该技术在娱乐视频与家庭监控业务中的典型应用场景,展望了超分辨重建技术未来可能的研究方向以及与相关图像处理技术相融合的发展趋势。

关键词: 超分辨率, 插值方法, 深度学习算法, 视频业务

Abstract:

The development history of super-resolution reconstruction technology and its typical approaches were briefly introduced.An implementation solution of super-resolution reconstruction was proposed on one kind of intelligent terminal device.The image super-resolution approaches implemented by interpolation algorithms and deep learning algorithms were experimented and simulated, and their results on terminal’s processing performance and images’ quality of experience were evaluated and analyzed.The suggestions on appropriate scenarios of super-resolution implemented by the intelligent terminal device were proposed.Furthermore, some typical cases of super-resolution reconstruction technology in the fields of entertainment video services and home surveillance services were discussed.The possible research direction and the trend of convergence of this technology and relevant image processing technologies were also prospected.

Key words: super resolution, interpolation approach, deep-learning algorithm, video service

中图分类号: 

No Suggested Reading articles found!