基于K-均值聚类的彩色图像质量评价及优化

针对彩色图像质量无法实时评价及优化的问题，提出了基于K-均值聚类的彩色图像质量评价及优化算法。首先利用K-均值聚类的方式构建样本数据集；然后通过待评价图像与聚类数据集之间的相似性来构建特征集；之后再将待优化图像与聚类数据集之间进行融合，对融合后的矩阵进行PCA变换实现了图像质量的优化；通过实验证明，所提评价算法与人眼主观视觉具有较好的一致性；同时，还能通过评价结果实现图像质量的自适应优化。

Color image quality assessment and optimization based on K-mean clustering

In view of the problem that the color image quality can''t be evaluated and optimized in real time, this paper proposed a color image quality assessment and optimization method based on K-means clustering. This algorithm used K-means clustering to build the sample set, and set up feature sets by the similarity between the image and the cluster data set. Then, it fused the image with the cluster data set and optimized the quality of the image by principal component analysis(PCA) transformation of the fused matrix. Experiments shows that the assessment algorithm has good consistency with human subjective vision, and at the same time, it can also achieve adaptive optimization of image quality through assessment results.