Kinasih, Citra Ayu Sekar (2021) Ekstraksi Data Bangunan Dari Data Citra Unmanned Aerial Vehicle Menggunakan Metode Convolutional Neural Networks (Cnn) Dan Object-Based Image Analysis (Obia) (Studi Kasus: Desa Campurejo, Kabupaten Gresik). Other thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
03311740000085-Undergraduate_Thesis.pdf - Accepted Version Download (4MB) | Preview |
Abstract
Seiring meningkatnya pembangunan akibat pertumbuhan penduduk, diperlukan suatu pengawasan dalam pemanfaatan lahan secara tepat salah satunya melalui pemetaan sebaran bangunan. Pemetaan sebaran bangunan dapat dilakukan dengan cara menganalisis citra penginderaan jauh yang diambil menggunakan berbagai wahana salah satunya menggunakan wahana Unmanned Aerial Vehicle (UAV) yang mampu menyediakan citra resolusi sangat tinggi. Namun, selama ini proses klasifikasi seringkali dilakukan dengan cara digitasi secara manual yang dianggap kurang efektif dan efisien sehingga dibutuhkan cara ekstraksi otomatis. Dalam penelitian ini metode Object- Based Image Analysis (OBIA) dan Convolutional Neural Networks (CNN) digunakan secara bersamaan untuk mengatasi tantangan ekstraksi bangunan menggunakan data citra foto udara resolusi tinggi pada Desa Campurejo, Kabupaten Gresik dengan menggunakan algoritma Mask R-CNN, di mana algoritma ini diharapkan mampu membantu proses perbaikan untuk hasil segmentasi yang tidak sempurna. Penggabungan antar kedua metode tersebut kemudian dibandingkan dengan hasil klasifikasi bangunan menggunakan metode Mask R-CNN. Hasil klasifikasi kemudian dilakukan validasi dan uji akurasi sehingga mampu menghasilkan peta sebaran bangunan skala besar yaitu 1:5000. Akurasi hasil klasifikasi bangunan dengan metode OBIA-MaskRCNN diuji dengan menggunakan confusion matrix yang menghasilkan nilai untuk Campurejo wilayah 1 precision 97,24%, recall 80,64% dan accuracy 78,84%. Sementara hasil metode OBIA-MaskRCNN untuk Campurejo wilayah 2 precision 92,46%, recall 89,01% dan accuracy 82,99%. Sebagai perbandingan dilakukan klasifikasi dengan metode Mask R-CNN yang memiliki precision 94,78%, recall 82,63% dan accuracy 79,03% untuk wilayah 1 dan precision 98,10%, recall 78,37% dan accuracy 77,20% untuk wilayah 2. Prosedur ini menunjukan potensi yang besar untuk memanfaatkan kombinasi Object-Based Image Analysis (OBIA) dengan Convolutional Neural Networks (CNN) dalam melakukan ekstraksi bangunan.
======================================================================================================================================
Along with increasing development due to population growth, a monitoring of land use is needed, which one of them is through mapping the distribution of buildings. Mapping the distribution of buildings can be done by analyzing remote sensing images taken using various vehicles, one of them is the Unmanned Aerial Vehicle (UAV) which capable of providing very high resolution images. However, up to now the classification process is often done by manual digitization which considered less effective and efficient so that an automatic extraction method is needed. In this study, the Object Based Image Analysis (OBIA) and Convolutional Neural Networks (CNN) methods were used simultaneously to overcome the problem in building data extraction using high resolution photo image data in Campurejo Village, Gresik Regency using the Mask R-CNN algorithm, where this algorithm is expected to be able to assist the improvement process for imperfect segmentation results. This combination from the two methods then compared with the results of building classification using the Mask R-CNN method. Then the classification results are validated and tested for accuracy to produce a large-scale building distribution map with scale 1: 5000. The accuracy of the results of building classification using the OBIA-MaskRCNN method was tested using a confusion matrix which resulted in a value for Campurejo region 1 with a precision of 97.24%, recall of 80.64% and accuracy of 78.84%. While the results of the OBIA-MaskRCNN method for Campurejo region 2, precision is 92.46%, recall is 89.01% and accuracy is 82.99%. As a comparison, classification is carried out using the Mask R-CNN method which has a precision of 94.78%, recall 82.63% and accuracy 79.03% for region 1 and precision 98.10%, recall 78.37% and accuracy 77.20%. for area 2. This procedure shows great potential to utilize the combination Object-Based Image Analysis (OBIA) with Convolutional Neural Networks (CNN) in extracting buildings.
| Item Type: | Thesis (Other) |
|---|---|
| Additional Information: | RSG 621.367 8 Kin e-1 2021 |
| Uncontrolled Keywords: | Bangunan, Convolutional Neural Networks, Mask RCNN, Object-Based Image Analysis, Foto Udara Building, Convolutional Neural Networks, Mask RCNN, Object-Based Image Analysis, Aerial Photography |
| Subjects: | T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques |
| Divisions: | Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geomatics Engineering |
| Depositing User: | Citra Ayu Sekar Kinasih |
| Date Deposited: | 14 Aug 2021 09:00 |
| Last Modified: | 16 Oct 2023 08:25 |
| URI: | http://repository.its.ac.id/id/eprint/86422 |
Actions (login required)
 |
View Item |