超高清内容清晰度用户体验质量评价

发布时间:2021-02-07 作者:朱文瀚,翟广涛,陶梅霞,杨小康,张文军 阅读量:
超高清内容清晰度用户体验质量评价 - 中兴通讯技术
您当前访问的的浏览器版本过低,为了给您带来更好的体验,建议您升级至Edge浏览器或者推荐使用Google浏览器
取消

超高清内容清晰度用户体验质量评价

作者:朱文瀚,翟广涛,陶梅霞,杨小康,张文军 阅读量:1195

超高清内容清晰度用户体验质量评价

朱文瀚,翟广涛,陶梅霞,杨小康,张文军
(上海交通大学,中国 上海 200240)

摘要:针对多媒体行业对超高清内容清晰度用户体验评价的迫切需求,提出了一种有效的无参考质量评价算法,以预测目标内容的用户感知体验,并区分原始和伪4K内容。通过对目标内容进行分割,利用局部方差选择了3个代表性子块代替全局来提高计算效率。针对超高清内容的特性,提取了复杂度特征、频域特征和像素统计特征。采用支持向量回归的方法将这些提取的特征融合为一个质量指标,以预测目标内容的质量分数。实验结果表明,本模型可以有效地评估用户感知体验,并具有良好的辨别真假4K内容的能力。 
关键词:用户体验质量;无参考质量评价;超高清;自由能原理;频域分析;自然图像统计 


Quality of Experience Estimation of Ultra High Definition Content

ZHU Wenhan, ZHAI Guangtao, TAO Meixia, YANG Xiaokang, ZHANG Wenjun
(Shanghai Jiao Tong University, Shanghai 200240, China)

Abstract: In response to the urgent demand for assessing the quality of experience of ultra high-definition content in multimedia industries, a no-reference quality assessment model is proposed to predict the perceptual quality of the target content and distinguish pristine 4K and pseudo 4K contents. Our model segments the image and chooses three representative patches by local variances to improve computing efficiency. According to the characteristics of ultra high-definition content, complexity features, frequency domain features and pixel statistics features are extracted from the representative patches. The support vector regressor is employed to aggregate these extracted features as an overall quality metric to predict the quality score of the target image. The experimental results demonstrate that the proposed method can effectively evaluate quality of user experience and has a great ability to distinguish true and pseudo 4K contents.
Keywords: quality of experience; no-reference quality assessment; ultra high-definition; free-energy principle; frequency domain analysis; natural scene statistics

在线PDF浏览:PDF
PDF