الفهرس 
ntroductionOnly 14 pages are availabe for public view
 |  | from | 95 |  |  |
| | | from | 95 | | |
Abstract
Satellite images have revolutionized environmental monitoring by pro- viding continuous access to observations of the earth. This presents a prediction model for mapping Land U use and Land Cover (LULC) us- ing multispectral satellite images with a spatial resolution of 3 meters, captured by a 4-bands PlanetScope satellite. The study employed 105 geo-referenced images categorized into eight different LULC classes and used various machine learning techniques such as Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), Normal Bayes (NB), and Artificial Neural Network (ANN) to train the pro- posed model on both raster and vector data. The proposed Land Use and Land Cover Raster Vector (LULCRV) model involves sev- eral phases, including pre-processing, labelling, model building, and prediction. Firstly, the preprocessing phase prepares the input data for the proposed LULCRV model, including a vector training set and raster mosaic images. It involves several steps, such as data collec- tion, data cleaning, and data integration. These steps are necessary to ensure that the input data is suitable for the subsequent phases of the LULCRV model. The preprocessing phase also includes im- age enhancement techniques to improve the quality of the input data. Secondly, the Labelling phase of the LULCRV model involves the extraction of objects from the input images based on their features, colors, shapes, and boundaries. This is accomplished through object detection algorithms such as thresholding, edge detection, and region growing. Once the objects have been detected, they are assigned a class number based on their characteristics and stored in a vector layer for further processing. Thirdly, the model building phase of the LULCRV model involves building the proposed LULC model using a set of classifier models, which includes SVM, RF, NB, DT, and ANN.