Abstract: Deriving multi-scale database from the third national land survey database is an important task for cartographers.In this paper, we focus on processing large number of Land polygons,study the raster generalization for land polygons to reduce broken and small land polygons, construct the semantic proximity model and the spatial geometric topology model,analyse land features map generalization based on the topology and geographical characteristics of the region, establish an automatic land generalization production process.Based on the above methods, the computer programming technology is used to develop land generalization system and generalize land features with about 230 000 land features of a county in Sichuan province to test the system.It turns out that the method proposed in this paper meets the requirements of thematic mapping and greatly improve the automatic generalization efficiency of land polygons.

Key words: multi-scale database, raster generalization for land polygons, the semantic proximity, spatial geometric topology model, generalization for land polygons