IMPROVEMENT OF SLOPE BASED FILTERING METHOD USING MULTIRESOLUTION TECHNIQUE

• Main Author: Amida Parvati, Gauri
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73546
• Institution: Institut Teknologi Bandung

The Digital Terrain Model (DTM) is a digital elevation model that has been augmented with elements such as breaklines observations in addition to the original data. One of the methods that has been used to filter the Digital Surface Model (DSM) into DTM is the Slope Based Filtering (SBF) method. This method performs filtering using distance and slope parameters. One of the drawbacks of the SBF method is that it is less effective when used to remove non-terrain objects that have dimensional differences. Because of this, the SBF method is developed using multi-resolution techniques. By using multi-resolution techniques, object data with large dimensions will be eliminated with low spatial resolution data while maintaining terrain data in high spatial resolution data. This study aimed to filter DTM using data with more than one resolution and combine them, producing DTM using three interpolation methods namely natural neighbor, kriging, and inverse distance weighted, as well as testing the accuracy of the resulting DTM. The data used in this study was DSM ITB Campus Jatinangor raster data taken using LiDAR. This research produced bare earth data resulting from multi-resolution processing as well as DTM from multi-resolution processing results. Then, from the results of the RMSE Z accuracy test, it was known that the manufacture of DTM using the natural neighbor method produced the better RMSE Z value compared to the other two methods, with an RMSE Z value of 0.49 m at the point-to-pixel accuracy test method and 1.034 m with the pixel-to-pixel accuracy test method.