聚类分析在高保真彩色印刷分色中的应用

Cluster analysis applied to separate colors in high-fidelity color printing

研究一种新的基于改进的K均值聚类的高保真彩色印刷分色方法,算法首先将彩色图像通过非线性变换,转换到Lab颜色空间,再利用改进的K均值聚类算法进行色彩学习,最后经过改进的误差分散求取结果。算法优点是利用图像的空间相关信息,使分色结果得到局部优化;同时,由于阈值的引入,可以很好地控制分类精度;最后,通过对误差分散算法的改进,同时保证了分色图像色彩的连续性与差异性。

A new modified K-means clustering-based high-fidelity color separation method was presented in this paper. In this algorithm, the authors transformed the image pixels from RGB color space to Lab color space by a nonlinear transformation, then after studying the samples of color information using modified K-means clustering and modified diffusing error, the authors made segmentation. The characteristic of this algorithm is using the spatial interrelated information, which makes the result of local optimization. At the same time, because of the introduction of the threshold, the authors can well control the classification accuracy. Finally, the continuity and color difference of color images can be ensured through modified error diffusion algorithm.