K-means algorithm is widely used in spatial clustering. It takes the mean value of each cluster centroid as the Heuristic information, so it has some disadvantages: sensitive to the initial centroid and instability. The improved clustering algorithm referred to the best clustering centriod which is searched during the optimization of clustering centroid. That increased the searching probability around the best centroid and improved the stability of the algorithm. The experiment on two groups of representative dataset proved that the improved K-means algorithm performs better in global searching and is less sensitive to the initial centroid.