We have implemented a semi-automatic
color segmentation scheme based on simulated annealing. Unlike automatic techniques,
which cannot know the region of interest to the user, this method allows a user
to very quickly identify an arbitrary but particular region and returns the
region more accurately segmented and appropriately measured.
This method can be used in a range of
applications such as image sub-selection in graphic arts, tumor identification
in medicine, or as an input mechanism for training samples in industrial
tracking. We have implemented this technique in an educational software tool
that allows students to take measurements on images and videos. Using the
region identification routine, students can compare the land areas of the
continents, the growth of the ozone hole, or the size of an enormous iceberg
which recently broke off an ice shelf in Antarctica.
 |
|
 |
|
Segmentation on image of iceberg. |
|
Segmentation of Australia.
|
 |
|
|
Segmentation result on image of ozone hole
from video sequence. |
|
|
|
