Classification of sample image

We have chosen one image to show how classification can be performed.

The software used for classification is Image Calculator by Ph.D Karsten
Heia (karsten.heia@fiskforsk.norut.no),

We have used uncalibrated data due to problems with the calibration lamp.
Shadows can be seen in both upper and lower part of the picture.
We hypothesize this can be due to distortion in the curved window onboard
the airplane or possibly because of non-uniform illumination of the entrance-slit.

Example of how to select classes.


Choosing representative parts of the image for every class.

After models for the classes have been calculated a classification has been
made. If the models are sufficiently accurate enough similar pictures can be
classified automatically.

Final product after classification. In the lower right corner misclassification
can be seen due to the above mentioned shadows.