基于改进K-means的压缩IP 由于k-means本身受异常点影响较大，这里采用迭代k-means的方法，降低异常点的影响，减少计算量和提高聚类数目的灵活性。并添加合并异常聚类方法，提高聚类的均匀性-K-means based on improved compression IP As k-means itself is influenced by outliers, where an iterative k-means method to reduce the impact of o