8/27/2009

MATRICES ADDITION
Given two matrices A and B if we want to add B to
A (that is form A+B) then if A is (nm), B must be
(nm), Otherwise, A+B is not defined.

2D TRANSFORMATIONS

The addition produces a result, C = A+B, with

elements:

Ci, j = Ai , j + Bi , j

Nihar Ranjan Roy

1 2 5 6 1 + 5 2 + 6 6 8
3 4 + 7 8 = 3 + 7 4 + 8 = 10 12

Nihar Ranjan Roy

MATRICES MULTIPLICATION (EXAMPL)

MATRICES MULTIPLICATION
Given two matrices A and B if we want to multiply
B by A (that is form AB) then if A is (nm), B must
be (mp), i.e., the number of columns in A must be
equal to the number of rows in B. Otherwise, AB is
not defined.

26+ 63+ 72=44

2 6 7 6 8 44 76
4 5 8 3 3 = 55 95

9 2 3 2 6 66 96

The multiplication produces

a result, C = AB, with
m
elements:
C = a b
i, j

6 8
2 6

4 5 3 3

2 6

ik kj

k =1

(Basically we multiply the first row of A with the

first column of B and put this in the c1,1 element of
C. And so on).
Nihar Ranjan Roy

Undefined!
2x2 x 3x2 2!=3

2x2 x 2x4 x 4x4 is allowed. Result is 2x4 matrix

Nihar Ranjan Roy

8/27/2009

MATRICES -- BASICS

TRANSFORMATION

Unlike scalar multiplication, AB BA

Matrix multiplication distributes over addition:
A(B+C) = AB + AC
Identity matrix for multiplication is defined as I.
The transpose of a matrix, A, is either denoted AT
or A is obtained by swapping the rows and
columns of A:
a1,1
A=
a2 ,1

a1, 2
a2, 2

a1,1
a1,3

A
'
=
a1, 2
a2,3
a1,3

Transformation is the process by which we can

change the shape , position and orientation of
an object
Basic

a2 ,1

a2, 2
a2, 3

Nihar Ranjan Roy

2D GEOMETRICAL TRANSFORMATIONS

Translate

Nihar Ranjan Roy

TYPES OF TRANSFORMATIONS
Geometric transformation when the object is moved
relative to a stationary coordinate system or
background it is called geometric transformation.

Shear

Coordinate Transformations in this transformation

the coordinate system is moved relative to the object
and the object is held stationary

Rotate Scale

Nihar Ranjan Roy

types of transformations are

Translation
Rotation
Scaling
Reflection
shearing

Nihar Ranjan Roy

8/27/2009

TRANSLATION

SCALING
We can specify how much bigger or smaller by means of a
scale factor

Translate the point p(x,y) to a new location which is given by

the vector v=txi+tyj;

To double the size of an object we use a scale factor of 2, to

half the size of an object we use a scale factor of 0.5

x' = x + t x
y' = y + t y

x' = s x x

In matrix format:

y' = s y y

x' x t x
y ' = y + t
y

If we define the translation matrix

P =P + T.

x' s x
y ' = 0

t x
T =
t y

, then we have
If we define

Nihar Ranjan Roy

ROTATION

sx 0
S sx , sy =
, then we have P =SP
0 sy
Nihar Ranjan Roy

10

MIRROR REFLECTION
Mirror reflection about x axis

Points can be rotated through an angle about the origin:

| OP ' |=| OP |= l

p`= M x ( p)

x ' 1 0 x
y ' = 0 1 X y

P(x,y)

x ' =| OP ' | cos( + ) = l cos( + )

= l cos cos l sin sin

= x cos y sin

y ' =| OP ' | sin( + ) = l sin( + )

= l cos sin + l sin cos

0 x
s y y

P(x,y)

x ' cos
y ' = sin

Mirror reflection about y axis

sin x
cos y

p`= M y ( p )

x ' 1 0 x
y ' = 0 1 X y

= x sin + y cos

P =RP
Nihar Ranjan Roy

11

Nihar Ranjan Roy

12

8/27/2009

SUMMERY

SHEARING TRANSFORMATION
The shear transformation cause the image to slant. X shear
maintains the y-coordinate but changes the x values which cause
the vertical lines tilt left to right

Shear in the x direction

SH

1
= 0

Simultaneous shearing

SH

a
1
0

xy

0
0

1
= b

Shear in the y direction

SH

a
1
0

Nihar Ranjan Roy

1
= b

0
1
0

0
0

0
0

Translate:
Scale:
Rotate:
Mirror
Shear

P = P+T
P = Ssx,sy(P)
P = R(P)
P = Mx(P) or P = My(P)
p=SHx(P) or p=SHx(P) or p=SHxy(P)

Multiplying versus adding matrix

Solution: Homogeneous Coordinates

Nihar Ranjan Roy

13

14

TRANSLATION AS HOMOGENEOUS
TRANSFORMATIONS

HOMOGENEOUS COORDINATES
For a given 2D coordinates (x, y), we introduce a third
dimension:
[x, y, 1]
In general, a homogeneous coordinates for a 2D point has
the form:
[wx, wy, w]

Now, redefine the translation by using homogeneous

coordinates:

Advantages

Similarly, we have:

matrix operations are simplified

Fractional calculations may be avoided example

tx x
ty y

1 1

0
1
0

P ' = Tv P

Scaling
x ' s x
y ' = 0

1 0

[1/2
1/3] in homogenous it can be represented as
[3 2 6]

P
Nihar Ranjan Roy

x ' 1
y ' = 0

1 0

x ' x t x
y ' = y + t
y

15

0
sy
0

Ssx,sy

Rotation
0 x
0 y

1 1

x ' cos
y ' = sin

1 0

Nihar Ranjan Roy

sin
cos
0

0 x
0 y

1 1

16

8/27/2009

ROTATION ABOUT AN ARBITRARY

POINT
Mirror Reflection about x-axis

p`= M x ( p)

& mirror reflection about y-axis

p`= M y ( p )

x' 1 0 0 x
y ' = 0 1 0 X y


z ' 0 0 1 z

x ' 1 0 0 x
y ' = 0 1 0 X y


z ' 0 0 1 z

R,p=Tv.R T-v

Simultaneous shearing
SH xy

1
= b
0

a
1
0

where

0
0
1

1 0 h
Tv = 0 1 k
0 0 1
17

Nihar Ranjan Roy

SCALING ABOUT AN ARBITRARY

POINT

1 0 h
Tv = 0 1 k
0 0 1

Ssx, sy

0
sy
0

Nihar Ranjan Roy

0
0
1

1 0 h
Tv = 0 1 k
0 0 1

Nihar Ranjan Roy

18

Let the line y=mx+c have a y intercept (0,c) and

Slope with x axis. The steps involved in this composite
Transformation are
1. Translate the intersection point A(0,c) to origin
2. Rotate by so that the line aligns with x axis
3. Mirror reflect about x axis
4. Rotate back by
5. Translate A back to point(0,c)

S(sx,sy),p=TV.Ssxsy .T-V
s x
= 0
0

cos sin 0
R = sin cos 0
0
0
1

REFLECTION ABOUT ANY LINE

Steps involved are

1. Translate the arbitrary point(h,k) to origin
2. Scale it with respect to origin
3. Translate back the arbitrary pt to original position
(inverse of step 1)
where

Let the arbitrary point be (h,k) and the point to be rotated

anti-clock wise is(x,y). The steps involved in transformation
are as follow
1. Translate (h,k) to origin
2. Rotate (x,y) by an angle
3. Translate the resulting point back to (h,k)

ML=TV .R .MX .R .T-V

1 0 h
Tv = 0 1 k
0 0 1
19

Nihar Ranjan Roy

20

8/27/2009

TRANSFORMATION ROUTINES
1 0
Tv = 0 1
0 0
1
Tv = 0
0

0
c
1

0 h
1 k
0 1

cos sin 0
cos sin 0
R = sin cos 0 R = sin cos 0
0

0
1
0
0
1
1 0 0
M x = 0 1 0
m
1
sin =
& cos =
0 0 1
m2 + 1
m2 + 1
finally

Slope m=tan() therefore

1 m 2
2
m +1
2m
ML = 2
m +1
0

2m
m2 + 1
m2 1
m2 + 1
0

2cm

m 2 + 1
2c
m 2 + 1
1

21

Nihar Ranjan Roy

IDENTITY-MATRIX(H)

ARGUMENT H is a transformation array of 3x3 elements

Local I,J variables for stepping through the H array
BEGIN
FOR I=0 TO 3 DO

FOR J=0 TO 3 DO
IF (I=J) THEN H[I,J]
1
ELSE
H[I,J]
0;
RETURN;
END

Nihar Ranjan Roy

22

TRANSFORMATION ROUTINES:-SCALE
MULTIPLY-IN-SCALE(SX,SY,H)
ARGUMENTS sx and sy scaling factor
H an identity matrix
LOCAL
I,J to iterate through matrix
BEGIN
FOR I=1 TO 3 DO
BEGIN
H[I,1]
H[I,1]*SX
S sx , sy
H[I,2]
H[I,2]*SY
END
RETURN
END

Nihar Ranjan Roy

Similarly write transformation routines for

Translation
Rotation

sx 0 0
= 0 sy 0

0 0 1

23

Nihar Ranjan Roy

24

8/27/2009

DISPLAY PROCEDURE

PROBLEM

The display procedure call involves

1. Saving the over all transformation matrix
2. Multiply the overall transformation matrix on the
left by the transformation in the call to form a new
overall transformation matrix
3. Transferring the control to the display procedure

Reflect the diamond shaped polygon whose vertices

are A(-1,0) B(0,-2) C(0,1) and D(0,2) about
1. Horizontal line y=2
2. Line y=x+2

A return from the display procedure involves the

following
1. Restoring the overall transformation matrix from
the saved value
2. Returning the control to the calling program
Nihar Ranjan Roy

25

Nihar Ranjan Roy

26

1) Since the line is parallel to x-axis we donot need to rotate the

polygon
A(-1,4)
2)

ML=TV MX T-V
B(0,6) C(1,4) D(0,2)
1 m 2
2
m +1
2m
ML = 2
m +1
0

A(-2,1) B(-4,2)

2m
m2 +1
m2 1
m2 +1
0

C(-2,3)

2cm

m 2 + 1 0 1 2
2c
= 1 0 2

m 2 + 1
1 0 0 1

D(0,2)
Nihar Ranjan Roy

27