基于双路编解码器的卷积神经网络在低照度图像增强和超分辨率中的研究  

Research on Convolutional Neural Network Based on Dual Codec in Low-light Image Enhancement and Super-resolution

作　　者：刘畅[1] 姜竹青[1] 李凯[1] Liu Chang;Jiang Zhuqing;Li Kai(School of Artificial Intelligence,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]北京邮电大学人工智能学院,北京100876

出　　处：《广播与电视技术》2022年第3期50-55,共6页Radio & TV Broadcast Engineering

基　　金：国家重点研发计划(2019YFC1511404);国家自然科学基金(62002026)资助。

摘　　要：本文以高清户外直播技术为切入点,针对夜晚场景下受光照条件限制直播画面亮度低、不清晰的问题,提出了一种基于双路编解码器的卷积神经网络,应用于低照度图像增强和超分辨率,有效提升了户外直播的画面质量,降低了对传输带宽的需求,有利于提升户外直播的用户体验。This paper takes HD outdoor live broadcast technology as research background.In night scenes,live broadcast videos captured under restriction of lighting conditions are unclear and have low brightness.To solve the problem,this paper proposes a convolutional neural network based on dual codec,which can be applied in low-light image enhancement and super-resolution.This network can effectively improve the quality of outdoor live broadcast in night scenes,reduce the demand for transmission bandwidth and help improve user experience of outdoor live broadcast.

关 键 词：低照度图像增强 超分辨率 卷积神经网络 双路编解码器 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]