Abstract: An automatic approach of updating vector edge in GIS based on Grouping Snake is proposed in this paper. The vector data existing in GIS are refined using the snake model in terms of information in remote sensing images, during the whole process of which, vector data act as both the updated objects, and the prior knowledge that conducts change detection in remote sensing images. There are two typical algorithms about snake model opposite in the computational complexity: the original snake model and the greedy algorithm. A trade-off Grouping Snake is put forward, attempting to group control nodes, in order to decrease the poor localization and the weak neighbour-node relationship, respectively caused by the original snake and the greedy algorithm. A few experiments are performed on vector edges and TM remote sensing image of Zhalong Wetland in China to validate Grouping Snake in vector edge update..