Joshua Dare-Cullen
Neural Compression for Hyperspectral Imaging
Background
BSc Computer Science; MSc Artificial Intelligence and Adaptive Systems
PhD
Neural Compression for Hyperspectral ImagingMy PhD focuses on developing new AI-based methods for compressing hyperspectral images, which are large 3D datasets with hundreds of spectral bands used in Earth observation, agriculture, and environment. I design neural compression models that combine deep learning, attention mechanisms, and entropy modeling, tailored specifically to the structure of hyperspectral data. The research is interdisciplinary, spanning AI, information theory, and remote sensing, and involves collaboration with industrial partners such as Ordnance Survey. In simple terms, I use AI to teach computers how to shrink very large satellite images while keeping the essential information, making them easier to store, transmit, and use for monitoring the environment and supporting decision-making.
Supervisors
-
Lead Supervisor
Professor Chunbo Luo
-
Lead Supervisor
Associate Professor Markus Mueller
External Partners
- Ordnance Survey