基于自然场景统计的无参考图像质量评价算法

为了度量多种失真类型的图像质量,提出一种基于图像空域自然场景统计特征的无参考图像质量评价算法。该算法通过度量失真图像和原始图像在统计规律上的偏差,对失真图像质量做出评价。与现有无参考图像质量评价算法相比,该算法不需要使用原始图像及其失真图像进行训练,也不需要知道图像的失真类型,是一种更具实际意义的通用型无参考图像质量评价算法。同时,考虑到人眼观察图像时感兴趣区域的影响,该算法加入了视觉显著性区域提取的过程。实验结果表明,该算法对于人的主观感知具有较好的一致性。

No-Reference Image Quality Assessment Algorithm Based On Natural Scene Statistic

To estimate a range of image distortions, a no-reference image quality assessment method is proposed based on the collection of statistical features from a space domain natural scene statistic (NSS) model. The proposed method makes an evaluation to the testing image, making use of measurable deviations from statistical regularities observed in natural images. Compared with existing no-reference image quality assessment methods, the proposed method does not require the training process between training images and human opinion scores, nor the prior knowledge of the type of distortions. Thus, it is a more significant universal no-reference image quality assessment algorithm. Simultaneously, the proposed method incorporates a salient region detection process, taking account the effect of the region of interest. The experimental results show that the proposed method is consistent with human subjective perception.