基于加权索引图的图像质量的半参考评测模型

Abstract

为了进一步提高半参考图像质量评价的准确性,设计了一种半参考图像质量评价方法,该方法建立传输前后图像特征的变化与图像主观质量之间的联系。选择映射机制与合理的特征,设计了能够表征特征变化情况和传输图像劣化位置的参考图像,在接收端对参考图像进行全参考质量评价就能判断特征变化情况。并引入两层HVS特性,分配图像不同分块的质量权重,将此质量权重作用在对参考图像的质量评价中,通过对参考图像进行基于视觉加权的全参考质量评价,来实现对传输图像的基于视觉的加权半参考质量评价。该方法比对参考图像采用传统的,不加权的PSnr方法以及SSIM方法在评测准确性上分别高出了18.65%,2.71%,单调性上分别高出17.86%,11.97%,一致性上分别高出55.18%,23.45%。

In order to further improve the accuracy of semi-reference image quality evaluation,a kind of semi-reference image quality evaluation method is presented,which establishes the relation between changes of image feature after transmission and subjective quality of image.In the quality evaluation system,mapping mechanism and rational characteristics are chosen.The reference image can represent the change of image feature and location of deterioration of the image.So the changes of feature can be figured out in receive end by full reference quality assessment of reference image.Two layers of HVS characteristics are used to distribute the quality weight of image blocks,which will be used in quality assessment of reference image.Through the full reference quality assessment based on the visual weight of reference image,the semi-reference image quality assessment of transmission image can be achieved.The evaluating accuracy of this method is respectively higher by 18.65% and 2.71% than psnr and ssim,while the monotonicity of this method is respectively higher by 17.86% and 11.97%,and the consistency is respectively higher by 55.18% and 23.45%.