适应于图像复原的无参考图像质量评价方法

No-reference Image Quality Assessment Metric for the Image Restoration

由于图像降质过程的复杂性、成像获取条件限制,以及图像本身的复杂性和图像复原过程的病态性,图像复原解大多都是近似的或畸变的,一种适应于图像复原质量评价的计算方法将大大提升图像复原的应用范围。针对图像复原过程的病态性,提出了一种针对图像复原图像质量评价的计算方法,该算法通过在图像质量算子中引入图像相似矩阵和图像复原趋势矩阵,使其能适应复原对于图像结构或噪声结构的变化。该图像质量评价算子计算无需参考图像,可以很好地反映图像的模糊程度和噪声程度,并且计算简单。实验证明了该图像质量评价算子的有效性。

Due to the complexity of image degradation, the restriction of the image capture, the complexity of images and the ill-posed problem of the image restoration process, the solutions of image restoration are almost the approximate or aberrant. An image quality assessment （IQA） method which is adaptive to the image restoration will advance the image restoration applications greatly. Aiming at the ill-posed problems of the image restoration, a no-reference IQA metric for the image restoration is proposed. The image similar matrix and the image restoration matrix are introduced in this metric, and it can adapt to the changes of image structure, noise structure or the noise quantity due to the image restoration. This metric can detect both blur and noise, and can be computed easily. The validity of the IQA metric is proved by experiments.