Hypersepectral Image Classification Using Hybrid Ensemble Elm

Abstract

Hyperspectral images (HSIs) are commonly utilized in remote sensing because of the continuity of the bands they contain. An important change in HSI categorization was brought about by the development of deep learning methods. The various Convolutional Neural Network (CNN) models are used in several HSI processing applications. The computational cost of HSIs is increased, and the Hughes effect is caused by, their higher dimensionality. As a result, dimensionality reduction (DR) is an essential preprocessing step for most CNN models. The incorporation of spatial and spectral data into HSI classification is a further obstacle to achieving reliable results. A few 3-D-CNN models are built to tackle this problem; however, they aren't as efficient as other approaches in terms of execution time. In this paper, we proposed a hybrid ensemble Extreme Learning Machine to capture spatial and spectral data from HSI images. Based on experimental results, the suggested ensemble model outperforms other models by 98%.