DEM derived from photogrammetric generated DSM using morphological filter

Digital Elevation Model (DEM) is a terrain model consisting of elevation data and representing the bare earth surface. Meanwhile Digital Surface Model (DSM) describes earth surface that includes nonterrain features such as vegetations and buildings. The most tedious and time consuming method of generating DEM is to gather ground points by conventional ground survey methods such as levelling or Global Positioning System (GPS) observations. An alternative approach is by photogrammetric technology on stereo pairs of aerial photographs. Measurement can be made either semi-automatic with human intervention or fully automatic taking advantage of the image matching algorithm. In built-up area, photogrammetric methods would generate DSM. DEM is produced by filtering out the nonterrain features. In this study the effect of various grid resolutions on the photogrammetric generated DSM and the morphological filtered DEM are investigated. It was observed that the height accuracy of photogrammetric generated DSM deteriorated rapidly above 1m grid resolution. The height Root Mean Square Error (RMSE) of the morphological filtered DEM obtained was 2.5m.