Image Segmentation Chin-Ya Huang Mon-Ju Wu

Image Segmentation
Chin-Ya Huang
Mon-Ju Wu
ECE 533 Final Project, Fall 2006
University of Wisconsin – Madison
Methodology
1) Acquire the color information and the
edge information separately.
2) Use hue, saturation and intensity to get
color information.
3) Use the Matlab “edge” command to
extract the image boundary.
4) Combine the above result by getting
the union of (2) and (3).
5) Final modification.
Methodology
Start
Original
Image
Separate image into
H. S. I components
Texture
Extraction
Extracted Edge
Intersaction
End
Shape
Extraction
Example
Target:
Extract the image
of the soccer
player out from
the entire image.
Step 1 : Using HSI as a threshold
A) Cut partial
images to
compute the HSI
values.
B) Set the threshold
value.
C) Examine through
the image to
acquire the color
information.
Step 2 : Using Matlab “edge” command
A) Use Matlab “edge” command to
extract the course edges.
B) Use Matlab “imfill” command to fill
the area whose surrounding course
edges make a closure.
edge
imfill(edge)
Step 3 : Combine the above images
A) Get the union of
the images from
Step 1 and Step 2.
B) Combine the
color information
and the edge.
Step 4 : Final modification
A) Perform dilate on
“edge” make sure
all the separate
line segments are
connected.
B) Use “imfill”
command to fill
the area inside
the boundary.
Step 4 : Final modification (cont.)
C) Intersect the
image from (B)
with the image
from Step 3 to
remove the noise.
D) Image from (C) is
the final result.
Final Result
Original Image
Future Work
A) Develop a more powerful algorithm to
perform interpolation in order to
connect the disconnected line
segment.
B) Find other ways to segment image
other than using HSI.
C) Develop a more powerful algorithm
for noise removal.