Abstract: This paper proposed a novel algorithm of color image segmentation, based on clustering and region merging. First, an image was smoothed while preserving the boundaries by mean shift algorithm. Second, the initial segmented regions were obtained using K-means clustering algorithm in the feature space. Finally, the initial regions were merged to form the final segmentation result by a new region merging strategy. The simulation results show that color image segmentation results of the proposed approach are well consistent with human perception.

Key words: color image segmentation, mean shift algorithm, K-means clustering, region merging

摘要： 提出一种基于K均值聚类与区域合并的彩色图像分割算法。首先，对图像运用mean shift算法进行滤波，在对图像进行平滑的同时保持图像的边缘;然后，运用K均值算法对图像在颜色空间进行聚类，得到初始分割的结果;最后，给出了一种区域合并策略，对初始分割获得的区域进行合并，得到最终的分割结果。仿真结果表明，算法的分割结果和人的主观视觉感知具有良好的一致性。

关键词: 彩色图像分割, 均值偏移算法, K均值聚类, 区域合并