Multispectral image segmentation utilizing constrained clustering approach and CGT classifier

The practical investigation on change detection (CD) on satellite data using machine learning techniques is the main emphasis of this work. Land plays a significant role in the procedure for categorizing and makes up a sizable portion of the Earth surface. High geographical and temporal resolution data are needed for the out coming parameters to be estimated. The accuracy of classification is significant in the analysis of change detection in multispectral images. The great deals of conventional algorithms that have been developed are unsuitable for typical color categorization since the color characteristic is not taken into account when creating the algorithm. Hence, a suitable method should be selected for the better segmentation and classification. The submitted approach provides a new advanced foundation for change detection and land cover classification. This study offers a new statistical framework for preparing multi-spectral remote sensing data based on preprocessing with an adapted homo-morphological filter (HMF) model. The extensive use of improving the convergence rate and detection performance is made by constrained proximal and fuzzy clustering segment (CPFCS) module. They are applicable to the examination of satellite image issues such as land cover categorization and monitoring of transformations using co-operative game theory (CGT). This classifier has generalization along with the most sensitive capability to initialization and pixel-level features. The clustering findings, as measured by several validity indices, demonstrate that the proposed algorithms produce higher-quality and more accurate clusters in land cover categorization and change detection challenges. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.