Klasifikasi Penutup Lahan Menggunakan DTM dan DSM LiDAR dengan Algoritma Support Vector Machine dan Random Forest (Studi Kasus: Institut Teknologi Sepuluh Nopember Kampus Sukolilo)

Nuraini, Annisa (2024) Klasifikasi Penutup Lahan Menggunakan DTM dan DSM LiDAR dengan Algoritma Support Vector Machine dan Random Forest (Studi Kasus: Institut Teknologi Sepuluh Nopember Kampus Sukolilo). Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 5016201057_Undergraduate_Thesis.pdf] Text
5016201057_Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2026.

Download (14MB) | Request a copy

Abstract

LiDAR adalah teknologi penginderaan jauh aktif yang memanfaatkan sinar laser untuk mendeteksi objek di permukaan bumi. Airborne LiDAR merupakan salah satu jenis lidar yang menggunakan wahana udara untuk proses penyiaman objek. Dari data LiDAR ini dapat diperoleh Digital Surface Model (DSM) yang selanjutnya dapat diekstrak menjadi Digital Terrain Model (DTM). Dalam perkembangannya untuk pengolahan data LiDAR, telah banyak digunakan perangkat lunak maupun dengan menggunakan algoritma yang dibangun seperti machine learning. Tujuan dari penelitian ini adalah memanfaatkan data LiDAR untuk klasifikasi penutup lahan dengan menggunakan machine learning, yaitu dengan algoritma Support Vector Machine (SVM) dan Random Forest (RF). Klasifikasi yang diterapkan adalah supervised classification dimana dibutuhkan data training untuk melakukan klasifikasi. Kelas penutup lahan yang diprediksi pada penelitian ini terbatas pada objek bangunan, vegetasi, jalan, lahan terbuka, dan badan air. Data yang digunakan untuk klasifikasi adalah data turunan dari LiDAR yaitu DTM dan DSM. Skema klasifikasi yang digunakan adalah dengan input satu data dan kombinasi data, serta diterapkan juga skema splitting ratio training point yaitu 70:30, 75:25, 80:20, dan 90:10. Hasilnya skema input satu data belum memberikan hasil yang optimal. Input kombinasi data memberikan penambahan akurasi dan mengasilkan akurasi yang baik yaitu lebih besar dari 0,80. Hasil terbaik didapat dari input kombinasi data DSM dan DTM rasio training testing 75:25. Pada metode SVM dihasilkan overall accuracy 0,824 dan kappa 0,780. Sedangkan pada metode RF dihasilkan overall accuracy 0,832 dan kappa 0,790. Secara keseluruhan, metode RF memiliki keunggulan dalam mengklasifikasikan objek bangunan dan lahan kosong, sedangkan metode SVM memiliki keunggulan dalam mengklasifikasikan objek jalan.
=================================================================================================================================
LiDAR is an active remote sensing technology that utilizes laser beams to detect objects on the earth's surface. Airborne LiDAR is one type of lidar that uses airborne vehicles for the object's illumination process. From this LiDAR data, a Digital Surface Model (DSM) can be obtained which can then be extracted into a Digital Terrain Model (DTM). In its development for LiDAR data processing, software has been widely used as well as using algorithms built such as machine learning. The purpose of this research is to utilize LiDAR data for land cover classification using machine learning, namely with Support Vector Machine (SVM) and Random Forest (RF) algorithms. The classification applied is supervised classification where training data is required to perform classification. The predicted land cover classes in this study are limited to building objects, vegetation, roads, open land, and water bodies. The data used for classification is derived from LiDAR data, namely DTM and DSM. The classification scheme used is a single data input and a combination of data, and the splitting ratio training point scheme is also applied, namely 70:30, 75:25, 80:20, and 90:10. The result is that the one data input scheme has not provided optimal results. Input data combination provides additional accuracy and produces good accuracy which is greater than 0.80. The best results are obtained from the input of a combination of DSM and DTM data with a training testing ratio of 75:25. In the SVM method, the overall accuracy is 0.824 and kappa is 0.780. While the RF method produced an overall accuracy of 0.832 and kappa 0.790. Overall, the RF method has an advantage in classifying building objects and vacant land, while the SVM method has an advantage in classifying road objects.

Item Type: Thesis (Other)
Uncontrolled Keywords: Classification, Land Cover, LiDAR, Random Forest, SVM, Klasifikasi, Tutupan Lahan
Subjects: G Geography. Anthropology. Recreation > G Geography (General) > G70.217 Geospatial data
G Geography. Anthropology. Recreation > G Geography (General) > G70.5.I4 Remote sensing
G Geography. Anthropology. Recreation > GA Mathematical geography. Cartography > GA102.4.R44 Cartography--Remote sensing
G Geography. Anthropology. Recreation > GA Mathematical geography. Cartography > GA105.3 Cartography.
G Geography. Anthropology. Recreation > GA Mathematical geography. Cartography > GA139 Digital Elevation Model (computer program)
Q Science > QA Mathematics > QA336 Artificial Intelligence
T Technology > T Technology (General) > T57.5 Data Processing
T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models.
Divisions: Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geomatics Engineering > 29202-(S1) Undergraduate Thesis
Depositing User: Annisa Nuraini
Date Deposited: 30 Jul 2024 01:18
Last Modified: 30 Jul 2024 01:18
URI: http://repository.its.ac.id/id/eprint/109916

Actions (login required)

View Item View Item