Abstract: Texture is one of the most important characteristics used in identifying objects or regions of interest in an image. Based on the fact that there is a high difference between a sharp image and its blurred version in texture features, a novel no-reference image blur metric based on the texture features is proposed. The key idea of the proposed method is to blur the original image with a weak and a strong low-pass filters, and then to compute the blur metric based on the texture feature variation between the two different blurred versions. The experimental results show that the proposed method is monotonic, robust to additive noises, and also consistent with human visual system.

Key words: no reference image quality assessment, blur measure, texture

摘要： 图像的纹理信息是人眼观察和识别物体的重要特征，根据清晰图像低通滤波前后纹理特征变化较大的特点，提出了基于纹理特征的无参考图像模糊度评价方法，对输入图像进行不同尺度的低通滤波，得到输入图像的两个副本，利用灰度共生矩阵求取两个副本图像的纹理特征变化量，以此作为图像的模糊评价依据。实验表明，该算法计算的模糊度相对于图像的模糊程度是单调的，具有良好的抗噪性，并且符合人眼视觉系统特性。

关键词: 无参考图像质量评价, 模糊度, 纹理特征