|
The
second and third figures show results obtained using segmentation with a
number of measures described in detail in the referenced paper, and using
one or more of the available polarisation images. Segmentation is used throughout
the study since it provides the greatest sensitivity for discriminating between
changes in image property because it achieves a filter window matched to
the homogeneous regions in the image. The best results are obtained using
all polarisations with the MTSEig procedure. Figure 4 treats these results
as "ground truth" and compares this with results obtained from segmenting
different polarisation measures.
- using all polarisations is the most important factor
- optimised polarisation measures, such as MTSEig and ML, outperform span by a factor of about two.
 |
Fig. 2
|
|
|
|
