COMPARATIVE STUDY OF FILTERING DIGITAL SURFACE MODEL TO DIGITAL TERRAIN MODEL USING SLOPE-BASED FILTERING AND GEODESIC MORPHOLOGICAL RECONSTRUCTION METHODS

• Main Author: Putri Adillah, Kurnia
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73413
• Institution: Institut Teknologi Bandung

The need for very high resolution spatial data is increasingly required by various institutions, both government and non-governmental organizations. Basic Geospatial Information in Indonesia such as base map, namely the Rupa Bumi Indonesia (RBI). One of the main information presented in the RBI Map is hypsographic information or information about height. Height data can be obtained from various sources. Terrestrial and extra-terrestrial measurement methods (Global Navigation Satellite Survey, GNSS) require a long time, while satellite imagery has limitations in spatial resolution so that it does not represent the situation in the field. Other technologies such as UAV- Photogrammetry and LiDAR can present the earth's surface with high spatial resolution. Digital surface data obtained from the acquisition of UAV-Photogrammetry still contains the height of objects above the earth's surface, to obtain the height of the ground surface requires a filtering process. To produce terrain height data that represents the actual situation, an automatic data filtering process is needed to get the best digital terrain results. In this final project, the process of filtering digital surface model data into digital terrain model data derived from aerial photography data will be carried out using two filtering methods, namely Slope-based Filtering (SBF) and Geodesic Morphological Reconstruction (GMR). This study aims to apply the filtering method commonly performed on LiDAR data to aerial photography data from UAV-Photogrammetry and analyze the results to find the best filtering method. The results showed that the SBF method performed better than the GMR method. The RMSE values for the open area samples, control points, and each pixel are 0.497 m, 0.355 m, and 1.771 m. Hopefully this final project can facilitate the extraction of information, especially height information on a large scale.