视频图像视觉兴趣区的提取及其在图像质量评估中的应用-估.PDF

视觉兴趣区的提取及其在视频图像质量评 估中的应用1 凌云 夏军 屠彦 尹涵春 （东南大学电子科学与工程学院，南京 210096） 【摘要】通过主观眼动跟踪实验和客观 Itti 模型分别分析了视频图像的感兴趣区域提取问 题。针对主观眼动跟踪实验，分析了眼动跟踪实验数据与视频图像的时间同步问题，并通过 确定注视点的范围得出视频图像兴趣区权重矩阵；针对客观Itti 模型，通过适当选取人眼在 观察视频时所能接受的感兴趣区域的个数，得出了视频图像的兴趣区权重矩阵。在此基础上， 分析和比较了利用主观眼动跟踪实验和客观Itti 模型得出的兴趣区权重矩阵对 PSNR 图像质 量算法性能的影响。实验证明，通过参数设置，主观眼动跟踪实验和客观Itti 模型提取的兴 趣区权重矩阵对 PSNR 都有明显的改善，改善后的模型不但保持了传统 PSNR 算法的简易 性，同时也提升了检测结果与主观感受的相关性。 【关键词】感兴趣区域；眼动跟踪；图像质量评价；PSNR 中图分类号：TN911.21 Detection of region of interest and application in video image quality assessment Ling Yun Xia Jun Tu Yan Yin Hanchun （School of Electronic Science and Engineering, Southeast University, Nanjing 210096, China ） 【Abstract 】The detection of region of interest in video content by the subjective eye-tracking experiment and Itti’s objective model is analyzed. For eye-tracking experiments, the time synchronization problem between video content and the data obtained by the eye-tracking experiments is discussed, and the saliency map of region of interest is determined by extension of the interesting point. The setting parameters of Itti’s model are discussed. The saliency map is formed by optimizing the number of interesting areas. Both subjective and objective models of region of interest are integrated into peak signal to noise ratio (PSNR) quality metric. The reliability improvement and difference are discussed. The experimental results show that by setting parameters, the application of the regions of interest obtained from both eye-tracking experiments and Itti’s model improves the correlation between the PSNR and subjective assessment while keeping the simplicity of objective quality assessment model. 【Key words 】region of interest; eye-tracking experiment; image quality assessment ; peak signal to n