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。
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 intel-ligent 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 su-per-resolution implemented by the intelligent terminal device were proposed.Furthermore,some typical cases of su-per-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.
作者
祝谷乔
姜超
徐煜烨
ZHU Guqiao;JIANG Chao;XU Yuye(E-surfing Digital Life Technology Co.,Ltd.,Shanghai 200085,China;China Information Communication Technologies Group Co.,Ltd.,Wuhan 430070,China)
出处
《电信科学》
2023年第7期156-165,共10页
Telecommunications Science
关键词
超分辨率
插值方法
深度学习算法
视频业务
super resolution
interpolation approach
deep-learning algorithm
video service