基于视觉分割图的图像质量评价方法.PDF

ISSN1009-3044 E-mail: eduf@ 第8 卷第7 期 (2012 年3 月) Computer Knowledge and Technology 电脑知识与技术 Computer Knowledge and Technology 电脑知识与技术 Vol.8, No.7, March 2012. Tel:+86-551-5690963 5690964 基于视觉分割图的图像质量评价方法 薛祥飞，桂小玲 （厦门大学计算机科学与技术系，福建厦门361005） 摘要：图像质量评价是图像处理领域的研究热点之一，传统的图像质量评价方法通常从图像的整体出发，没有充分考虑到人眼视觉 对图像的感知过程，往往造成质量评价的结果与人眼视觉机制的感知质量不符。该文利用Itti 人类视觉注意模型，得到图像的显著 点，再利用分割算法得到分割图以及分割区域的权值对图像进行质量评价计算，在LIVE 图像库上的实验结果表明，该文的方法的 评价结果相比于MSE(Mean Square Error) ，PSNR(Peak Signal Noise Ratio)方法与图像的主观评价（由DMOS 值表示）结果更一致。 关键词：图像质量评价；视觉注意模型；分割图；MSE ；PSNR 中图分类号：TP391 文献标识码：A 文章编号：1009-3044(2012)07-1601-03 ImageQualityAssessmentBasedonVisualSegmentationMap XUE Xiang-fei, GUI Xiao-lin (Department of Computer Science, Xiamen University, Xiamen 361005, China) Abstract: Image quality assessment is one of hot research fields of image processing. The traditional method of image quality assessment starts from the overall image, which does not fully take the process of image visual perception into account. And it always causes the evalua⁃ tion result of the conventional algorithms does not match the ones of human vision. In this paper, the Itti human visual attention model is taken to get the image salient points, then use the segmentation map which getting from some segmentation algorithms, and use the right value of segmented region to calculate the image quality assessment. The experiment in LIVE image database show that, the method given by this paper is closer to image subjective evaluation presented by DMOS value, than the MSE (Mean Square Error) and PSNR (Peak Sig⁃ nal Noise Ratio). Keywords:image quality assessment; visual attention model; segmentation map; MSE; PSN