利用图像复原产生全参考质量评价所需的参考图像.文中提出一种新颖的含噪音图像的无参考质量评价算法,利用小波去噪算法获得近似的参考图像,通过人眼视觉系统处理,计算待测图像与参考图像的峰值信噪比.LIVE图像数据库上,文中提出的算法以及NPSNR和HVSPSNR与主观感知质量的相关系数分别是93.02％,91.29％和90.62％.实验结果表明,该方法的性能表现要优于传统的NPSNR和HVSPSNR.%This research aims at generating reference images required by full-reference image quality assessment by using the image restoration algorithm. This paper proposes a new algorithm for no-reference image quality assessment of noisy images. An approximation to reference images is obtained by using an image denoising algorithm, then the peak signal noise ratio of the approximated reference to the distorted image is measured after they are processed by the human vision system model. On the test platform of LIVE image database, the correlation coefficients between subjective ratings and objective scores measured by the proposed method, NPSNR, HVSPSNR are 93. 02%, 91. 29% and 90. 62%, respectively. The experimental results show that the proposed method performs better than the traditional NPSNR and HVSPSNR.
展开▼
