利用图像二阶差分信息的图像质量评价方法

Quality assessment method based on second order differential of image

客观图像评价指标广泛应用于图像处理领域,不同的评价指标在不同失真类型图像中的适用性存在较大的差别。梯度结构相似性指数(GSSIM)利用图像一阶差分信息进行评价,在应用中存在一些不足。图像的二阶差分信息不仅反映了图像的结构信息,而且能够响应图像梯度的变化。将图像二阶差分信息、对比度、亮度相结合,提出基于图像二阶差分信息的图像质量评价方法DG-SSIM,实验结果表明,DG-SSIM相对于其他评价指标能够更好地符合人眼视觉系统(HVS)的特性,且图像质量评价模型的性能指标也优于其他评价指标。

Objective evaluation is widely used in the field of image processes. There are big differences between the different evaluation indexes over different types of image distortion. Gradient structure similarity index （GSSIM） using image information of a first-order differential evaluation has some deficiencies in the application. Second derivative information of images reflects not only the structural information, but also the changes in the image gradients. The image quality evaluation method combined with the second derivative image information, contrast, and brightness, named as DGSSIM, is proposed, and the experimental results show that DGSSIM can meet the human visual system （HVS） ＇better, and its image quality and performance evaluation model are also better than the other evaluations.