Clusters in Hyper-Cubic Multi-Channel Satellite Imagery

Kenneth A. Hawick



Multi-spectral remotely-sensed data such as satellite imagery can yield excellent insights into complex phenomena such as weather systems. Analysing the multi-channel space to separate out different features still presents a challenge, which may increase with the availability of hyper-spectral satellites. We use component labelling and population thresholding techniques to separate out clusters in hyper-dimensional channel space and use this information to identify different cloud types in geostationary satellite imagery. Three dimensional visualisation techniques are used to study the hyper-dimensional channel population data.

Important Links:

Go Back