基于梯度及HVS特性的离焦模糊图像质量评价

Defocused image quality assessment based on gradient and HVS

在当前人眼视觉系统(HVS)特性研究的基础上提出基于梯度及HVS特性的离焦模糊图像质量评价模型(GVSSIM)。该模型利用Sobel边缘梯度算子提取图像的梯度信息,并根据人眼视觉特性进行视觉加权,得到新的结构相似性评价指标,进而获得图像质量评价指标。该方法与SSIM(图像结构相似度)评价模型相比,具有计算简单的特点,对离焦模糊图像的评价结果能更好地反映人眼视觉感受。

Based on the latest study on human visual system（HVS） characteristics,this paper proposed the assessment model of the defocused image quality based on the gradient and HVS characteristics （GVSSIM）. This model took use of the Sobel edge gradient operator to extract the gradient information of the image, and weighed the formula in accordance with the characteristics of HVS to get the similarit assessment indicators of the new structure, and accessed to the image quality assessment indicators. This method is characteristics of simple computation and the assessment result of the defocused image better reflect the human subjective feeling compared with the SSIM evaluation model.