This paper presents a method of clustering which improves the stability and efficiency of K-means clustering and based on density.The first step of this method is to select the largest density point of the data set .On this basis, the farthest point from this point is selected as the second initial cluster center,Then, we look for the farthest distance from the two initial points in the remaining points as the center of the third clusters,And so on, until you find the desired K points, and then according to the K-means algorithm to update the cluster center, until convergence or Up to the set number of iterations. The result of the experiments show that the proposed method is better than the traditional K-means method in terms of accuracy and stability, and can be used as a new idea of clustering research.