基于切比雪夫矩无参考模糊图像质量评价

提出了一种新的基于Tchebichef矩的无参考模糊图像质量评价方法。将模糊图像通过低通滤波得到再模糊图像；将模糊图像和再模糊图像分别进行8×8分块并计算每一图像块的Tchebichef矩；根据Tchebichef矩块的值将原始图像块分为平滑块，纹理块和边缘块，计算原始图像和再模糊图像对应块之间的Tchebichef矩向量相似度，得到三类图像块的局部平均相似度；进行融合得到原始图像的最终评价质量。实验结果表明，该方法优于其他算法，与主观评分有更好的一致性，能够更准确地评价模糊图像质量。

No-reference blur image quality metric based on Tchebichef moment

A novel no-reference image quality assessment metric based on Tchebichef moment is proposed for blurred images. The re-blurred image is produced by blurring the original blurred image with a low pass filter. Then the original blurred image and re-blurred image are divided into 8 × 8 blocks respectively. Tchebichef moment of each block is computed. The original blurred image blocks are classified into smooth block, texture block and edge block according to the Tchebichef moment block value. The local quality index is obtained from the similarity measure of moment vectors between the re-blurred and the original image blocks. The local quality indices of three types are rolled as the final image quality. Experimental results show that the proposed metric is better than other methods. It obtains high correlations with subjective quality evaluations. So the proposed metric is appropriate for no-reference blurred image quality assessment.