Role of Convolutional Neural Networks in Hyperspectral Imaging Applications: A Review

In last few years, considerable improvements have been made in digital image acquisition and processing, improving image quality and results. However, hyperspectral imaging techniques evolved to gain more valuable information from images that integrate digital spectroscopy and imaging to generate information in special and spectral dimensions. HSI is a powerful tool for image analysis in potential application areas, including remote sensing, agriculture, horticulture, food quality assessment, medical imaging, forensic science, etc. Artificial intelligence technologies are used as an important tool in many research areas. As a result, image analysis and classification studies are carried out on HSI with various ML techniques. Lately, deep learning with convolutional neural networks has emerged as an important tool for dealing with hyperspectral images’ high dimensionality. In this paper, we review work done on different hyperspectral imaging applications using convolutional neural networks, which shows that deep learning using convolutional neural networks can be a powerful tool for hyperspectral image analysis and classification.