Algorithm development for hyperspectral image classification

Abstract

Recent advancements in remote sensing technology have significantly enhanced our ability to monitor and interpret the Earth's surface with high precision and consistency. Among these advancements, the shift from multispectral to hyperspectral imaging marks a pivotal transformation. Hyperspectral imaging, with its hundreds of contiguous and narrow spectral bands, enables more accurate material and terrain classification due to its superior spectral resolution. In this work, we explore the potential of deep learning-based methods to classify hyperspectral images by leveraging both spatial and spectral information. Our study implements and compares traditional 2D Convolutional Neural Networks (2D CNNs), which focus on spatial feature extraction, and 3D Convolutional Neural Networks (3D CNNs), which simultaneously capture spatial and spectral correlations, offering a more holistic representation of hyperspectral data.