On the basis of analyzing the Support Vector Machine (SVM) classification technique and the features of LIDAR data and aerial imagery，this paper has put forward a new building extraction method based on object-oriented SVM，which integrates multi-source information of aerial imagery and Light Detection and Ranging (LIDAR) data. Tests show that the extraction accuracy is improved by using this method. Moreover，the proposed object-oriented building extraction method not only proves to be effective but also can update GIS database quickly and accurately.