CLASSIFICATION METHODS FOR REMOTELY SENSED DATA: LAND USE AND LAND COVER CLASSIFICATION USING VARIOUS COMBINATIONS OF BANDS
| Autor: | Jasmee Jaafar, Nur Anis Mahmon, Azita Laily Yusof, Norsuzila Ya'acob |
|---|---|
| Rok vydání: | 2015 |
| Předmět: | |
| Zdroj: | Jurnal Teknologi. 74 |
| ISSN: | 2180-3722 0127-9696 |
| DOI: | 10.11113/jt.v74.4838 |
| Popis: | Land use and land cover (LU/LC) classification of remotely sensed data is an important field of research by which it is commonly used in remote sensing applications. In this study, the different types of classification techniques were compared using different RGB band combinations for classifying several satellite images of some parts of Selangor, Malaysia. For this objective, the classification was made using Landsat 8 satellite images and the Erdas Imagine software as the image processing package. From the classification output, the accuracy assessment and kappa statistic were evaluated to get the most accurate classifier. Optimal performance was identified by validating the classification results with ground truth data. From the results of the classified images, the Maximum Likelihood technique (overall accuracy 82.5%) was the highest and most applicable for satellite image classifications as compared with Mahalanobis Distance and Minimum Distance. Whereas for land use and land cover mapping, the RGB 4, 3, 2 band combinations were found to be more reliable. An accurate classification can produce a correct LU/LC map that can be used for various purposes. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|