Master’s Thesis titled: Implementation of Satellite Images Based on Clustering and Framework

The Faculty of Computer Science and Mathematics, Department of Computer Science, discussed the master’s thesis titled “Implementation of Satellite Images Based on Clustering and Framework” by the student Riyad Abdul Amir Ghafil. The thesis aims to study four stages: data collection, Hadoop clustering, satellite image classification, and accuracy assessment. In the first stage, data is acquired (i.e. data collection) and processed into a format that can be processed by Hadoop using ArcGIS. The second stage is preparing the infrastructure to install Hadoop, such as the virtual system and the Ubuntu operating system. The third stage is to group similar regions according to the number of centers specified in the K-Means algorithm. In the fourth stage, these areas collected from the previous stage are classified and their accuracy is calculated in the ArcGIS program, where the results appear in the form of four categories: (water, vegetation, bare lands, and buildings). It can be seen that the area has been secured before. Experimental results of the classification method showed sufficient performance. The analysis compares the accuracy of two clustering algorithms, ISO Clustering and K-Means. Evaluation metrics like Kappa percentage and F-score percentage are used to assess their performance. The average overall accuracy, F-score, and Kappa of ISO are 77.860%, 70.479%, and 75.599%. The thesis was successfully accepted.