Hyperspectral Image Intelligent Processing and Application

Abstract : Hyperspectral imaging is a powerful technique capable of obtaining both spatial and spectral information from a target by combining conventional machine vision and point spectroscopy methods. With the rapid development of satellite and unmanned aerial vehicle technology, hyperspectral images have witnessed a huge growth in data and have found numerous successful applications in daily life. Regarding the massive amount of hyperspectral data available, the need for methods to process and interpret hyperspectral images automatically, efficiently, and accurately, presents a significant challenge in the research and application of hyperspectral images. Recently, artificial intelligence technologies, such as machine learning and deep learning, develop prosperously, showing a promising perspective in overcoming the challenges in hyperspectral image processing. This talk presents a comprehensive overview of intelligent hyperspectral image processing and an application of hyperspectral images in tree species identification. First, we give a brief introduction to the characteristics and applications of hyperspectral images. Then, we introduce some classic and advanced techniques for the intelligent processing of hyperspectral images, including techniques for hyperspectral image enhancement, feature extraction, and interpretation. Finally, we present an application of hyperspectral images with multi-source data such as multispectral images, unmanned aerial vehicle LiDAR data, backpack LiDAR data, and in-situ data, in tree species identification, which is of vital importance for accurate calculation of carbon sink in urban areas.