基于局部结构张量的无参考型图像质量评价方法

A No-reference Image Quality Assessment Method Using Local Structure Tensor

该文提出一种基于局部结构张量奇异值分解的无参考型图像质量评价方法,由于图像局部结构张量能反映图像几何结构,因此利用张量特征值之间的关系来度量图像噪声与模糊水平,将两个度量结合得到图像质量的综合评价。通过分析仿真图像和实际图像的质量评价结果,该方法能同时度量因噪声和模糊造成失真后的图像质量。与图像质量评价数据库的主观评价结果比较表明,该文方法与主观评价结果相关性强,能很好地反映图像质量的视觉感知效果,并且易于实现。

A new image quality metric is proposed, it can be used to predict the no-reference image quality. Based on the Singular Value Decomposition （SVD） of the local structure tensor of the image, the noise and blur level is measured using the characteristic of singular value. The performance of the method is evaluated with a publicly available database of images and their quality score. The results show that the proposed no-reference method for the quality prediction of noise and blur images has a comparable performance to the leading metrics available in literature, and also that the inethod is easier to implement.