Combiner of classifiers using Genetic Algorithm for classification of remote sensed hyperspectral images

In the past few years, hyperspectral images have been considered as one of the most important tool in land cover classification due to its capability to obtain rich information of materials on earth surface. In this work we aim to produce an accurate thematic map for the remote sensed hyperspectral image classification problem, which is obtained using a combination of several classification methods. Three types of feature representation and two learning algorithms (Support Vector Machines (SVM) and Backpropagation Multilayer Perceptron Neural Network (MLP)) were used yielding six classification methods to perform the combination. Our combination proposal is based on Weighted Linear Combination (WLC), in which weights are found using a Genetic Algorithm (GA) - WLC-GA. Experiments were carried out with two well-known datasets: Indian Pines and Pavia University, and we observed that our proposed WLC-GA method achieves the highest accuracy among traditional Conscious Combiners, the widely used Majority Vote (MV) and Weighted Majority Vote (WMV), for both datasets.