关键词: 图像分割, 不完全K均值聚类, 分类优化

Abstract: To improve the clustering efficiency and image segmentation effect, the paper proposed an Incomplete K-means and Category Optimization (IKCO) method. First of all, the algorithm used simple approach to finish data subsampling and initial centers determining. Then, according to the clustering rules, the proposed algorithm finished image's segmentation. Finally, the algorithm used category optimization method to improve segmentation results. The experimental results show that, compared with the traditional K-means clustering method, the proposed algorithm has better segmentation efficiency, and the segmentation result has a higher consistency with human visual perception.

Key words: image segmentation, incomplete K-means clustering, category optimization