Image processing methods such as atmospheric correction in blue electromagnetic regions is major problem at present in this field. The problem got widen in case of space based data... Simultaneously the problem associated with getting an easy method for dimensionality reduction have also wast scope.
Thank you for your inputs. I am hoping to work on hyperspectral images for my PhD. Will get back to you if I have any specific questions. Thanks again.
Anomaly detection still remains a problem, most methods of data processing do not take into account that noise is signal dependent and of different intensity in component images. This can be topic for studying with particular applications.
The best application of Hyperspectral images is Satellite Images or aerial images.
If you can add optimisation techniques along with Image processing algorithms the
Research in Hyperspectral images will be fruitful for your PhD work, if any details required about it you can contact me here to discuss about future works.