Remote sensing image classification using multidimensional wolf pack optimization-based support vector machine classifier

The categorization of remote sensing (RS) images is an integral part of the earth observation technology that relies on RS data and has found widespread use in both the military and civilian sectors. The high complexity and limited availability of labeled samples in RS data, however, make executing RS image classification difficult from a scientific and practical perspective. With the advent of cutting-edge machine learning (ML) methods in recent years, ML-based approaches to RS image categorization have made substantial advances, opening up new avenues of inquiry and growth. Thus, in this study, we provide a novel ML method called multidimensional wolf pack optimization-based support vector machine (MWPO-SVM). The satellite images are first acquired and then preprocessed using the Adaptive Contrast Enhancement Algorithm (ACEA) and a median filter. Then, principal component analysis is utilized to leverage low-dimensional, more expressive characteristics to minimize the dimensionality of satellite images. Finally, satellite image categorization is carried out using the suggested approach. Extensive tests of the settings of remote sensing image classification are carried out on three various conventional methods to verify the efficacy and superiority of our suggested method. According to experimental findings, the suggested technique, when compared to conventional methods, generates more accurate classifications in a shorter amount of computing time.