|
|
Joint segmentation versus 3x3 average |
 |
 |
Finally, we compare
the classification result for joint segmentation and
classification with that for joint classification of the
averages over a 3x3 window. The latter clearly shows many
small errors. The average probability of correct
classification is 0.883 for joint segmentation and
classification compared with 0.738 for joint
classification of the 3x3 average |
Joint
segmentation
|
3x3
average
|
|
|
|
Reference:
T. Macri
Pellizzeri, P. Lombardo, C.J. Oliver
A new maximum likelihood
classification technique for multitemporal SAR and multiband optical
images
IEE Int. Symposium IGARSS, 2002, Toronto, Canada, June 2002.
|
|
