KLASIFIKASI PENUTUP LAHAN MENGGUNAKAN DATA LIDAR DENGAN PENDEKATAN MACHINE LEARNING

Abstract

Lidar is a remote sensing technology. Lidar data is widely used and has been developed for mapping, detailed spatial planning, and natural disaster analysis. In its development for Lidar data management, software applications are widely used as well as by using built algorithms such as machine learning. The research aims to utilize Lidar data for land cover classification using machine learning, namely Support Vector Machine (SVM). The research location is Tanjung Karang village, Mataram City, Lombok. The classification applied is a supervised classification in which the training data is needed to perform the classification. The predicted land cover class in this study is limited to buildings, vegetation, roads, open land. The data used for classification is derived from Lidar, namely DTM, DSM, nDSM, and Intensity. The classification scheme used is one data input and a combination of data. The reference data used is a topographic map (Topographic map of Indonesia). The results showed that the classification with a data combination scheme had a better accuracy value than the one data classification scheme, which increased accuracy by about 15-20%. This shows that there are complementary factors between the data to be able to identify objects in the classification process.