Text

Semantic interpretation of dense urban areas of Indonesia using various classification methods

---

Airbone LIDAR data classification using machine learning algorithm has been a hot topic within the geospasial information community. LIDAR data interpretation based on point clouds is a major domain of automatic mapping. With the rapid development of urban areas, updating database of the geospasial information is urgently needed. Usually, updating of the geospatial information use manual interpretation which is timw consuming and costly. In order to find the best method as well as to improve the efficiency, an automatic approach of data interpretation is absolutely needed.
However, each study area has specific feature characteristics and it can influence the accuracy of the classification results of the urban landscape. For this reason , this thesis explores appropriate LIDAR data classification methods that can be used in Indonesia. The rsearch structure consistt of 4 components: types of neighborhood, feature extraction, feature selection assessment, amd classification. To obtain optimal results, we used the comparison of quantitative results with different parameter configurations and evaluation using reference labels with respect to the semantic classes (Natural Ground, Building, Artificial Ground and Vegetation).
Promosing result of accuracy were obtained, 95.72% for combination of Support Vector Machine (SVM) algorith, K-nearest neighborhood with radius 20 neighbors (K20) and 3D-Eig + H group features, then 95,28% for combination of Suport Vector MAchine (SVM) algorith, spherical neoghborhood with radius 2m (S2) and all features, and 95,04% for combination of random forest (RF), cylindrical neighborhood with radius 3m(3m) and all features. Moreover the quantitative comparison result confirms the effectivenes of the combination method of selection of types of environment, multi-source features, and classification algorithm.

---

Ketersediaan

B20190725314

621.3678 TEG s

Perpustakaan BIG (600)

Tersedia

Informasi Detail

Judul Seri

-

No. Panggil

621.3678 TEG s

Penerbit

Berlin : Technische Universit Berlin., 2018

Deskripsi Fisik

48 hlm. : illus. ; 30 cm.

Bahasa

Inggris

ISBN/ISSN

-

Klasifikasi

621.3678

Tipe Isi

text

Tipe Media

other

Tipe Pembawa

unspecified

Edisi

-

Subjek

Penginderaan Jauh

Info Detail Spesifik

-

Pernyataan Tanggungjawab

-

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar