Classification Of Multi-Spectral Satellite Image Using Hierarchical Clustering Algorithms

This work presents hierarchical clustering algorithms for solving the task of crop classification using a multispectral satellite image. The hierarchical clustering algorithms uses splitting and merging techniques, w here splitting is used to obtain ideal possible clusters along with its centers. The clusters with its centers are then merged based on a parametric method. Three hierarchical clustering algorithms, namely, the Niche Hierarchical Artificial Immune System (NHAIS), Niche Particle Swarm Optimization (NPSO), and Niche Genetic Algorithm (NGA) are applied here for classification. To demonstrate the robustness of the proposed algorithm, results are presented for a real-time multispectral satellite image and an additional benchmark data set from the University of California, Irvine (UCI) repository. A performance comparison between all the three hierarchical clustering algorithms are presented and analyzed. The obtained results show that the NHAIS is most efficient among presented approaches.