摘要
已有的基于误差统计量的评价算法存在误判的缺点,因此提出基于结构相似性的全参考图像质量评价算法。算法通过实际图像和原始参考图像的亮度项、对比度项、结构相似度项三个方面信息的比较,评价失真图像的质量。实验结果表明,提出的算法优于MSE和PSNR。
Existing error statistics based algorithms have a disadvantage of erroneous judgment, so a full reference image quality assessment algorithm based on SSIM is proposed to overcome the disadvantage. The computation of luminance term contrast term and structure term in both actual image and original reference image is performed and gives a quality assessment value. Experimental results show that the proposed algorithm outperform both MSE and PSNR algorithms.
出处
《信息通信》
2013年第1期1-2,共2页
Information & Communications
