Contour line generation with point cloud filtering techniques

Three dimensional (3D) point cloud data are point data that contain the coordinate and height values of the objects on the surface. Today, cloud production has become very easy and its usage area has become widespread thanks to LiDAR systems and developing optical- photogrammetric software. In this thesis, contour line generation in forest areas with point cloud filtering methods for small scale photogrammetric maps. For this 3D point cloud data was produced by Semi Global Matching (SGM) algorithm from aerial photographs which taken by digital air camera, in three different areas selected as low, medium and high density forest area and medium rugged terain. After the forest areas in this data were filtered with spectral values of points and Cloths Simulation Filter (CSF) algorithm which is available in a open source software . Later, Digital Terrain Surface was produced from this filtered data through Triangular Irregular Network (TIN), Inverse Distance Weighting (IDW) and Kriging methods. Looking at the results obtained, filtering the wooded areas in the point cloud data with the spectral values of the points was more successful than filtering with the CSF algorithm and it was seen that the digital terrain model produced by TIN interpolation reflects the reference digital terrain model better than produced by IDW and Kriging methods. In addition, when the horizontal and vertical square mean errors of this contour lines were examined remain within the limiting values in the low and medium density woodlands. Limit values determined according to the slope values of the land. However, the horizontal and vertical accuracy of this contour lines mean square error exceed the limit values in the high density woodland.