Biomedical Engineering Reference
In-Depth Information
ii.
Update the contour information
(b)
Iterations++;
7.
End and fit spline to the final contour.
Note that in both cases it important to ensure that any movement of the contour
that violates the topological circle configuration of the curve is avoided, that is,
there cannot be any self-intersecting lines in the contour.
APPENDIX B
Incorporation of a priori information in the form of some probabilistic ap-
proach is detailed in this appendix. Specifically, the pseudocode for defining
intensity-based object background confidence classification is provided.
Pseudocode forObject-BackgroundClassificationBasedon Intensity
1.
Input Image
I
(
x, y
)
2.
Input object distribution information (object mean, object std devia-
tion)
3.
Input background intensity distribution (background mean, back-
ground std. deviation)
4.
Construct One-Sided Gaussian Probability Density Function:
(a)
If (Object mean
>
Background mean)
i.
For (intensity =Minimum Intensity; intensity
<
=Maximum
Intensity, intensity++)
A.
if (intensity
<
=Object Mean)
B.
Object Probability (intensity) = exp(-(intensity - Object
Mean)
2
/(2*(Object Std Deviation)
2
);
C.
else
D.
Object Probability (intensity) = 1.0;
E.
if (intensity
<
=background mean)
F.
Background Probability (intensity) =1.0;
G.
else
H.
Background Probability (intensity) = exp(-(intensity -
Background Mean)
2
/(2*(Background Std Deviation)
2
);
(b)
else
i.
For (intensity =Minimum Intensity; intensity
<
=Maximum
Intensity, intensity++)
