基于结构相似度的彩色图像质量评价算法

目的研究在Matlab中度量彩色图像之间结构相似度的评价算法,用以评估彩色图像的质量。方法以灰度图像的结构相似度评价模型(SSIM)为基础,在Matlab中将以RGB色空间存储的彩色图像转换到亮度信息和色彩信息分别度量的YUV色彩空间中,设计算法度量彩色图像之间的亮度比较值、对比度比较值和结构比较值,综合三部分后可得彩色图像之间的结构相似度。结果实验结果表明,利用彩色结构相似度算法可以计算彩色图像与参考图像之间的相似度数值,用作质量评价,且其评价结果与人类视觉感知评价(即主观评价)结果保持一致。结论该方法可作为一种新型的彩色图像质量评价准则。

Color Image Quality Assessment Algorithm Based on Structure Similarity

The work aims to study the assessment algorithm of structural similarity between color images in the Matlab to evaluate the quality of color images. Based on the structural similarity index measurement (SSIM) of the grayscale image, in Matlab, the color images stored in RGB color space were converted into the YUV color space in which the brightness and color were measured respectively. Assessment algorithm was designed to measure values of the luminance comparison, the contrast comparison and the structure comparison. The structure similarity between color images could be acquired by integrating the three parts. The experimental results showed that this algorithm of color structural similarity could calculate the degree of similarity between color images and reference images and be used for quality evaluation. Moreover, the evaluation results are consistent with the evaluation of human visual perception (i.e. subjective evaluation). This method can be used as a new type of color image quality evaluation criteria.