Abstract: A novel split-and-merge method with ranking selection （RSM） is proposed for the polygonal approximation of curves.We apply the ranking selection scheme of genetic algorithm to the split-and-merge process and substantially reduce the sensitivity of the traditional split-and-merge method to the initialization solution.The experimental results show that RSM is robust and outperforms the traditional split-and-merge method and the genetic-algorithm.We also apply RSM to the polygonal approximation of the satellite image of lake and obtain satisfactory results.