With the development of sensors like digital aerial cameras and remote sensing and photogrammetry technologies, such as oblique photography and three-dimensional technology, higher resolution and richer data are available for urban land cover extraction, including digital orthophoto maps (DOMs), true digital orthophoto maps (TDOMs), and topographic information. On the basis of highly overlapped oblique photogrammetry, the output of TDOM can solve most of the occlusion problems caused by illumination and observation angles. Making full use of spatial information, especially height, may reduce spectral confusions among urban land covers due to limited spectral resolution in RGB images. We acquired and processed information from TDOM and height data based on aerial sensors and oblique photogrammetry and used a multi-level threshold method combining the vegetation color index, brightness, spatial information, and height to rapidly extract urban land cover information. The method offers a simple and effective way to automatically extract complex land cover information on megacities with no need for training sample selection. This method was evaluated by comparing its results from the area of Beijing with those from the most widely used support vector machine (SVM) and maximum likelihood classification methods through visual inspection and quantitative measures, and this method achieved a better accuracy.