DESIGN OF CURVES 107
′≡
′
′
⎡
⎣
⎢
⎤
⎦
⎥
⎡
⎣
⎢
⎤
⎦
⎥ ′≡
′
′
⎡
⎣
⎢
⎤
⎦
⎥
⎡
⎣
⎢
⎤
⎦
A ⎥
x
y
xp
yq
B
x
y
xp
yq
=
and =
1
1
1
1
2
2
2
2
Let ′
′
′
⎡
⎣
⎢
⎤
⎦
⎥
′
′
⎡
⎣
⎢
⎤
⎦
C ⎥
x
y
x
y
* = +^1
1
2
2
λμ
=
+
=
+ – ( + )
+ – ( + )
1
1
2
2
12
12
λμ
λμ λμ
λμ λμ
xp
yq
xp
yq
xx p
yy q
⎡
⎣
⎢
⎤
⎦
⎥
⎡
⎣
⎢
⎤
⎦
⎥
⎡
⎣
⎢
⎤
⎦
⎥
=
{ + – } – ( + – 1)
{ + – } – ( + – 1)
12
12
λμ λμ
λμ λμ
xxp p
yyq q
⎡
⎣
⎢
⎤
⎦
⎥
Like points A and B, if C was expressed in the new coordinate system without any change in its
position due to the changed axes, then
′
⎡
⎣
⎢
⎤
⎦
C ⎥
xxp
yyq
=
+ –
+ –
12
12
λμ
λμ
Note that C′ and C′* are not identical, and there is a change in the position of C due to the shift in
the origin O to O′ by (p,q). To make this transformation also affine, the arbitrary scalars λ and μ need
to be constrained as
λ + μ – 1 = 0 or λ + μ = 1 (4.39)
Affine combination is therefore a type of linear combination wherein the respective weights sum
to unity. It is needed to preserve the relative positions of points (describing a geometric entity) during
transformation (change of coordinae axes), which is ensured by Bernstein polynomials in case of
Bézier curves.
4.4 Bézier Segments
Forn + 1 data points Pi,i = 0, ... , n, a Bézier segment is defined as their weighted linear combination
using Bernstein polynomials (Eq. (4.33))
r( ) = uCuu Bu ui=0 (1 – )– = =0 ( ) , 0 1
n
n i ni ii
i
n
i
ΣΣPPn i ≤≤ (4.40a)
y′ y
y′
A C
B
(OC = λOA + μOB)
x′
θ
θ
O x
(a) (b)
y
A C
B
(OC = λOA + μOB)
o′
O x
x′
Figure 4.15 Affine transformations and relation between weights λλλλλ and μ