CD Skripsi

Penerapan Teknik Multi-Temporal Cloud Removalpada Citra Satelit Untuk Segmentasi Badan Air Menggunakan Arsitektur Watnet

---

Satellite imagery is important in environmental monitoring, especially in water resources mapping and management. However, the presence of clouds is often a barrier to obtaining high-quality imagery, which can reduce the accuracy of water body segmentation. This research developed the Multi-Temporal Cloud Removal (MCR) method to overcome the cloud cover problem in Sentinel-2 satellite images. This MCR method utilizes temporal information from a series of images to replace cloud-covered areas with data from cloud-free images. This results in more precise and accurate images for water body segmentation. This study tested the MCR images using the WatNet segmentation model, a deep-learning architecture designed for water surface mapping. This research also modifies the cloud masking process by adding a dilation operation to detect all cloud-contaminated areas properly. The results showed that the MCR method developed was able to improve the quality of satellite images by significantly reducing cloud cover, with a Normalized Root Mean Square Error (NRMSE) value of 0.029 and a Peak Signal- to-Noise Ratio (PSNR) of 31.26 dB. Evaluation of the segmentation performance using WatNet showed an improvement in the accuracy of water body detection in the MCR-processed image compared to the cloudy image. The WatNet model tested on the MCR image achieved an accuracy value of 0.95, a dice coefficient of 0.92, and an Intersection over Union (IoU) of 0.89. These results indicate that the application of MCR can effectively improve image quality for water body segmentation analysis, enabling more precise and efficient monitoring of water resources.
Keywords: Multi-Temporal Cloud Removal, Satellite Imagery, Water Body Segmentation, WatNet.

---

Ketersediaan

Informasi Detail

Judul Seri

-

No. Panggil

2103112616

Penerbit

Pekanbaru : Universitas Riau FMIPA Sistem Informasi., 2024

Deskripsi Fisik

-

Bahasa

Indonesia

ISBN/ISSN

-

Klasifikasi

2103112616

Tipe Isi

-

Tipe Media

-

Tipe Pembawa

-

Edisi

-

Subjek

SISTEM INFORMASI

Info Detail Spesifik

-

Pernyataan Tanggungjawab

Mutia

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar