基于低阶Gaussian-Hermite矩的图像质量评价  

Image Quality Assessment Based on Low Order Gaussian-Hermite Moments

作　　者：唐腾健[1] 朱玥[1] 

机构地区：[1]重庆工业职业技术学院招生就业处,重庆401120

出　　处：《重庆师范大学学报（自然科学版）》2016年第3期126-131,共6页Journal of Chongqing Normal University:Natural Science

摘　　要：提出一种基于低阶Gaussian-Hermite矩的图像全参考质量评价方法。正交矩在图像处理和模式识别方面有很重要的作用,低阶矩可以从图像的分块中提取重要特征,这些特征通过计算两幅图像间相对矩误差得到,这种差异的大小更能体现出失真图像块相对原始图像块的失真大小,所以能更好的描述图像的质量。利用Gaussian-Hermite矩在一些公开的数据库中对提出的方法进行验证,实验结果和对比表明所提出的方法具有很好的效果。This paper presents a new full reference image quality metric based on Gaussian-Hermite moments.Orthogonal moments are powerful tools in pattern recognition and image processing applications.Low order moments are employed to extract important features in a block manner,from the image.The sign of the moment is taken into account to generate the relative moment difference,which is obtained by comparing the absolute moment difference to the moment of the reference image.The relative moment difference would reflect more accurately distortion between distorted image block and original image block.So this will have more ability to describe the image quality.The performance of the proposed method is evaluated on several public databases using GaussianHermite moments.Experimental results and comparisons demonstrate the efficiency of the method.

关 键 词：GAUSSIAN-HERMITE矩 图像全参考质量评价方法 相对矩误差 绝对矩误差 

分 类 号：TN911.73[电子电信—通信与信息系统]