Abstract: The characteristices of farmland site are complicated and versatile affected by many factors in urban fringe. Extract farmland and its corridor landscape types using high resolution remote sensing image is a superior way to detect farmland change. It is significant for farmland protection and delineation of basic farmland to analyze the relationship between farmland and its corridor landscape types. Taking Daxing district of Beijing as a case, object-oriented classification approach was adopted to extract farmland and other landscape types from SPOT5 remote sensing image in year of 2004 and 2007, including transport land, river, hydraulic construction land, irrigational land, farmland shelterbelt. The extracted road width was also evaluated. Finally, quantitative changes and spatial distribution changes of the 5 corridors during 4 years simply were analyzed through methods of corridor analysis of landscape ecology and GIS spatial analysis. The results showed that the classification accuracy and Kappa coefficient were more than 90% with visual interpretation, and the width of road was from 4 to 70 m; Farmland shelterbelt suffered the most obvious quantitative changes, and became scattered. Agricultural irrigational land distributed evenly. Secondary class road took greater changes, which mainly distributed in the northern study area affected greatly by urbanization and industrialization. Therefore, the object-oriented classification approach can take full advantage of class-related features, which fuses class site information at different levels. The spatial distribution of corridors and farmland shows a certain correlation, that is, the higher density of farmland shelterbelt and agricultural irrigational land, the smaller loss probability of farmland; the higher density of transport land, the higher loss probability of farmland.