Slides of review talk " The Challenges in Spectral Image Analysis: an Introduction, and Review of ANN Approaches", introducing a special session on "Spectral Image analyses with ANNs", organized by E. Merényi, at ESANN'99, European Symposium on Artificial Neural Networks, Bruges, Belgium, April 21-23, 1999.  See Paper for text and references. 

Slide 1


Slide 2

Concept of (hyper)spectral imaging.  From Campbell, 1996.


Slide 3

The difference between traditional broad band and hyperspectral imaging: bandpasses of Landsat TM (at bottom), and AVIRIS (top).


Slide 4

Hyperspectral (AVIRIS) image of Mammoth Mountans, California. 1024 x 512 pixles, 224 bands, approximately 300 Mbytes od data. 


Slide 5

Another illustration of the difference between broad band multi-spectral and hyperspectral resolution of the spectral signatures of surface materials. From Clark, 1999.



Slide 6

Methematical model for spectral data.



Slide 7



Slide 8



Slide 9



Slide 10

Example of the advantage of Self-Oragnizing Artificial Neural Net for clustering a spectral image. For more explanation of this slide  see AISRP'99  Workshop presentation.



Slide 12



Slide 13



Slide 14



Slide 15



Slide 16