基于自然场景统计的图像质量评价算法  

Image quality assessment algorithm based on natural scene statistics

作　　者：秦小倩 杜浩 QIN Xiaoqian;DU Hao(Guilin University,Guilin 541006,China;Guilin University of Electronic Technology,Guilin 541004,China)

机构地区：[1]桂林学院,广西桂林541006 [2]桂林电子科技大学,广西桂林541004

出　　处：《现代电子技术》2023年第23期36-42,共7页Modern Electronics Technique

基　　金：桂林学院2022年校级中青年教师科研基础能力提升项目:无参考图像质量评估研究;上海市在线检测与控制技术重点实验室开放基金项目(ZX2021104)。

摘　　要：基于自然场景统计(NSS)提出一种无参考图像的图像质量评估(IQA)算法,该算法的实现结合了图像的小波能量差异和人类视觉系统(HVS)的对比度敏感功能(CSF)模型。自然图像经过小波多尺度变换后的能量呈线性分布,而图像失真破坏此线性关系。所提算法利用图像失真这一自然特性的变化建立预测模型,预测失真图像各尺度的小波子带能量,并用失真图像在各尺度能量的预测值和实测值的差异度量失真图像的质量,再通过CSF模型进行优化调整,使其更接近人类视觉系统。该算法在LIVE图像质量评价数据库和CSIQ图像质量评价数据库进行测试,结果表明该算法与现有无参考图像质量评价算法相比,与人眼具有良好一致性、运行时间短的优势。A no⁃reference image quality assessment(IQA)algorithm is proposed based on natural scene statistics(NSS),which combines wavelet energy difference and contrast sensitive function(CSF)model of human visual system(HVS).The energy distribution of natural images after wavelet multi⁃scale transformation is linear.However,image distortion destroys this linear relationship.In the proposed algorithm,a prediction model is established by using the change of the natural characteristics of image distortion,and the wavelet sub⁃band energy of each scale of the distorted image is predicted,and the quality of the distorted image is measured by the difference between the predicted value and the measured value of the energy in each scale.And then,by the CSF model,it is optimized and adjusted to make it closer to the human visual system.The algorithm is tested in the image quality evaluation database LIVE and CSIQ.The results show that the algorithm has the advantages of good consistency with human eyes and short running time in comparison with the existing non⁃reference image quality evaluation algorithm.

关 键 词：图像质量评价 NR-IQA算法 小波变换 自然场景统计 HVS CSF模型 

分 类 号：TN911.73-34[电子电信—通信与信息系统]