基于最优聚类数和直方图匹配的图像彩色化方法

Colorization method with optional number of cluster and histogram matching

针对颜色转移彩色化算法中速度慢、效果不佳及人工干预性强等问题,提出了一种新型的彩色化算法。该算法首先解决了图像聚类中聚类数的选取问题,然后利用聚类算法分别对目标图像和源图像进行聚类分割,之后用直方图匹配算法使各目标图像块自动找到匹配的源图像块,将源图像块的颜色转移到目标图像块中,实现局部图像彩色化,最后合并各结果图像。实验结果表明,该算法比以前算法在彩色化的速度和质量上有较大改进,且自动化程度高。

With regard to the low speed, unsatisfactory results and high manual intervention of color-transfer colorization algorithm, a new colorization algorithm was introduced. Firstly the problem of clustering number selection was resolved during image clustering, then object image and source image were segmented respectively using clustering algorithms. After that, histogram matching algorithm was adopted for object block images. To achieve local images colorization, the color was transferred from source block images to object block images. Finally, colorized block images were merged. The experimental results show that the algorithm is better than the previous in terms of speed, quality and automation of colorization.