The Region Growing technique for Image segmentation
was implemented using the following procedure -
1. The user is asked to mark a seed point form where
the region growing should initiate.
2. Then the user is asked to enter an Intensity Threshold
to be used in determining connectivity of neighbouring points.
3. Then the N4 neighbours of the seed are checked
to see if they are within the threshold specified i.e. are they N4 connected
to the seed. If yes then they are marked
and then theri neighbours are checked and son on till no more N$ connected
points
can be found.
The various results obtained are shown below
|
|
|
|
Seed at (12,203) |
Seed at (41,155) |
|
|
|
Seed at (30,89) |
Salient Features of this Implementation :
1. Highly efficient due to use of a FIFO Queue data
structure -
All the points which are N4 connected
to a particular point are entered into the Queue, and the routine runs
till the Queue
becomes empty.
2. A check is maintained that no point is ever checked
twice for connectivity by using a systemm of tagging the points which have
been checked once.
2. Very Interactive - allows the user to vary a lot
of parameters at runtime such as -
a) The Interactive marking of
the seed point using a Java program.
b) The choice of using an old
seed file if present.
c) The Intensity Threshold used
to check connectivity.
Page last updated on 28 January, 2004. | AT cse.iitd.ac.in | © Parag Chaudhuri , 2009 |