Mohd Noor, Muhammad Fareed and Rahmat, Noor Zarikh Khuzaimi and Abdullah, Nur Azam and Mohd Ibrahim, Azhar and Al Mahmud, Suaib and Mardzuki, Muhammad Imran
(2025)
Machine learning-based change detection for land use land cover in Malaysia.
Mekatronika : Journal of Intelligent Manufacturing and Mechatronics, 7 (2).
pp. 124-133.
E-ISSN 2637-0883
Abstract
Remote sensing has gained widespread attention due to its applications and technological advancements. This study aims to explore the use of remote sensing for change detection for land use land cover (LULC). The study begins by focusing on pre-processing, including radiometric, geometric, and atmospheric correction as well as image enhancement to produce quality images for further classification analysis. Two classification methods were explored: supervised and unsupervised. For supervised classification, Support Vector Machine (SVM), Classification and Regression Tree (CART), and Random Forest classifiers were tested. After thorough evaluation, it was determined that the Random Forest algorithm was the optimal choice, yielding a training accuracy of 99.6% and a test accuracy of 80% for LULC classification. For unsupervised classification, a cluster classifier was used. Change detection is then conducted through image subtraction of two different timelines. Supervised classification of LULC images resulted in a total change of 94.74 km² across three locations: Wang Kelian, Sungai Golok and Pengerang while unsupervised classification resulted in change of 23.56 km² for Lahad Datu and 20.51 km² for Sungai Golok
Actions (login required)
 |
View Item |
