基于掩盖效应的无参考数字图像质量评价

针对数字图像的客观质量评估，从噪声检测的角度，提出了一种基于掩盖效应的无参考图像质量评估方法。首先对Hosaka分块进行了改进，取消了该方法对图像尺寸的限制。通过分块，将图像以不同的频率成分区分开来。然后计算各个子块的噪声。根据图像的污染程度，提出了基于掩盖效应的无参考图像峰值信噪比NPSNR。实验结果表明，该方法具有无参考、运算复杂度低、主客观较一致等特点。

No-reference digital image quality evaluation based on perceptual masking

From the angle of noise detection, this paper presents a mask-based no-reference approach of image quality assessment. Firstly, the approach improved Hosaka plots method and wiped out the limits on image size. Then this improved method was applied to the image to make plots and modified directional gradients were employed to abstract sub-blocks' noises. Finally, a new image PSNR was computed according to the pollution degree. Experimental results show that this no-reference approach can not only make the subjective assessment in line with the objective one, but also feature a low computational load.