基于FSIMc的图像质量评价模型  

Image quality assessment based on FSIMc

作　　者：许俐 姜秀华[1] XU Li;JIANG Xiu-hua(School of Information and Communication Engineering,Communication University of China,Beijing,100024)

机构地区：[1]中国传媒大学信息与通信工程学院

出　　处：《中国传媒大学学报（自然科学版）》2019年第5期60-65,共6页Journal of Communication University of China：Science and Technology

摘　　要：视觉是人们认知接受信息的主要渠道,是人类认知客观世界最主要的途径。图像在制作、压缩编码和传输的整个过程中,不可避免地会引入失真。图像质量评价模型从客观上用人类视觉感知系统的各种已知属性来量化失真图像和参考图像之间损伤的可见性。与主观评价方法相比,客观评价具有速度快、费用低、易于实现、自动实时监控等优点。本文在FSIMc基础上考虑颜色损伤的视觉特性,利用饱和度感知对算法进行改进。实验结果表明,本文模型能够提升算法性能,实现和人眼主观感知较好的一致性。Vision is the main channel for people to recognize and accept information,and it is the most important way to understand the objective world.The image inevitably introduces distortion throughout the production,compression coding,and transmission.The image quality assessment model objectively quantifies the visibility of damage between the distorted image and the reference image using various known properties of the human visual perception system.Compared with subjective evaluation methods,objective evaluation has the advantages of high speed,low cost,easy implementation,and automatic real-time monitoring.In this paper,the visual characteristics of color damage are considered on the basis of FSIMc,and the algorithm is improved by using saturation perception.The experimental results show that the proposed model can improve the performance of the algorithm and achieve better consistency with the human subjective perception.

关 键 词：图像质量评价 客观算法 图像颜色分析 视觉感知特性 

分 类 号：TN919.8[电子电信—通信与信息系统]