基于自适应查找表的高动态范围内容映射算法  

High Dynamic Range Content Inverse Tone-mapping Operator Based on Adaptive Look-up Table

作　　者：程宇红[1] 郭铖 CHENG Yuhong;GUO Cheng(Hunan Institute of Science and Technology,Yueyang 414006,China;State Key Laboratory of Media Convergence and Communication,Communication University of China,Beijing 100024,China)

机构地区：[1]湖南理工学院,湖南岳阳414006 [2]中国传媒大学媒体融合与传播国家重点实验室,北京100024

出　　处：《湖南理工学院学报（自然科学版）》2023年第4期19-23,共5页Journal of Hunan Institute of Science and Technology(Natural Sciences)

摘　　要：高动态范围是媒体行业超高清标准体系的重要组成部分.随着标准的完善、政策的支持和消费市场的演进,高动态范围行业的矛盾转移到内容上,需要能将现有普通图像或视频上转换为高动态范围显示的算法,该过程称作逆色调映射,是非适定的底层视觉问题,故通常结合深度学习方法来解决.将深度学习与图像转换或映射的常用高效方法(自适应查找表)结合,提出面向用户终端的高效逆色调映射算子,并针对高动态范围容器更高量化位深的特性进行了优化.实验结果表明,该方法相比其他基于深度学习的逆色调映射算法在表现与计算开销上均有优势.High dynamic range(HDR)is an important part of media’s ultra-high definition(UHD)standard system.With the recent advances in standardization,policy and consumer market,the contradiction of HDR industry comes to the aspect of content where conventional footage needs to be up-converted to HDR for display.This process is called inverse tone-mapping(ITM),an ill-posed low-level vision task,thus researchers are involving deep learning for implementation.This paper combines deep learning with self-adaptive look-up table(LUT)which is a popular and efficient method for image conversion,and proposes a user-end-oriented efficient ITM operator which is further optimized for higher-bit-depth HDR processing.The experimental results show that this method has advantages over other deep learning-based ITM operators on both performance and computational overhead.

关 键 词：查找表 高动态范围 逆色调映射 深度学习 

分 类 号：TN919.8[电子电信—通信与信息系统]