Biomedical Engineering Reference
In-Depth Information
T
X
=
{
x(0), x(1), x(2),.....,x(N-1)
}
(a)
Construct the coordinatematrix:
T
Y
=
{
y(0), y(1), y(2),.....,y(N-1)
}
(b)
Constructmatrix fromexternal energy and information:
B
x
(
•
,t
−
∆
t, t
−
2∆
t
)
,
B
y
(
•
,t
−
∆
t, t
−
2∆
t
)
, where
B
x
denotes the
x
-
derivative values of the external energy
∂x
E
ext
and
B
y
=
∂
∂y
E
ext
.
t
represents the iteration index with separation
∆
t
=1
for most cases.
(c)
Solve
X
=
K
−
1
B
x
and
Y
=
K
−
1
B
y
(d)
iterations++
(e)
ReSample Curve
2.
End and fit spline to the final contour.
B. Pseudocode for Snake ComputationUsing Local Neighborhood
Energy-Based Approach
1.
Input image
I
(
x, y
)
2.
Preprocessing
(a)
Compute feature map (Input Image
I
)
i.
Compute gradient map G =
∇
G
σ
⊕
I
A.
Normalize G
ii.
Compute regional information-based normalized feature
map F
3.
Input discrete points
4.
Define a contour through the sample points on the curve
5.
Input parameter values for snake computation:
α
= strength of elas-
ticity;
β
= rigidity strength;
γ
= gradient strength;
η
= other factor
(might
be regional or user-defined constraints). The number of parameters
introduced will be equal to the number of force fields used.
6.
While (Iterations != Maximum Iterations)
(a)
For (i=0; i
<
N; i++) /* N = Number Of Control Points */
i.
Search the
3
×
3
neighborhood (for 2D)
A.
Compute Energy due to all the factors at each neighbor-
hood
B.
Find the lowest energy neighborhood pixel
C.
Assign the point to this lowest energy neighborhood