显著性增益的块视差立体图像质量评价方法  

Saliency-Gained Block Parallax Stereoscopic Image Quality Assessment

作　　者：徐世稳 谷潇[1,2] 周军 XU Shi-wen;GU Xiao;ZHOU Jun(Institute of Image Communication ＆ Network Engineering,Shanghai Jiao Tong University,Shanghai 200240,China;Key Laboratory of Digital Media Processing and Transform,Shanghai 200240,China)

机构地区：[1]上海交通大学图像通信与网络工程研究所,上海200240 [2]上海市数字媒体处理与传输重点实验室,上海200240

出　　处：《电脑与信息技术》2018年第6期15-17,23,共4页Computer and Information Technology

摘　　要：随着立体影像内容的产出与消费体量爆发式增长,立体影像的舒适度问题引起越来越多的关注。文章结合块视差立体图像质量评价方法,提出了一种基于时序感知的图像显著性感知模型,并用图像显著性对块视差图进行增益,从而获得增益后的图像质量评价模型。实验结果表明,该方法较块视差立体图像质量评价模型的准确性有所提升。With the explosive growth in the output and consumption of stereoscopic image content, the comfort ofstereoscopic images has attracted more and more attention. In this paper, based on the block parallax stereo image qualityassessment method, an image saliency perception model based on temporal perception is proposed. And then this model isapplied to gaining the block parallax map to obtain a saliency-gained stereo image quality assessment model. Theexperimental results show that the proposed method improves the accuracy of the block parallax stereo image qualityassessment model.

关 键 词：立体图像质量评价 图像显著性 块视差 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]