Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Alternative Definitions:
t
j 2 f ( )
Gx ,1 t , f e ( t ) 2
e 2
x d
normalization
Gx ,2 t , f 2 e
4 ( t )2 j 2 f
e x d
Gx ,3 t , e ( t ) 2 / 2 j
e x d
( t ) 2 t
1 j ( )
Gx ,4 t , e 2
e 2
x d
2
2
Main Reference
S. Qian and D. Chen, Sections 3-2 ~ 3-6 in Joint Time-Frequency
Analysis: Methods and Applications, Prentice-Hall, 1996.
Other References
D. Gabor, Theory of communication, J. Inst. Elec. Eng., vol. 93, pp.
429-457, Nov. 1946. ( Gabor transform)
M. J. Bastiaans, Gabors expansion of a signal into Gaussian elementary
signals, Proc. IEEE, vol. 68, pp. 594-598, 1980.
R. L. Allen and D. W. Mills, Signal Analysis: Time, Frequency, Scale, and
Structure, Wiley- Interscience.
S. C. Pei and J. J. Ding, Relations between Gabor transforms and
fractional Fourier transforms and their applications for signal processing,
IEEE Trans. Signal Processing, vol. 55, no. 10, pp. 4839-4850, Oct. 2007.
3
( t )2 t
1 t 4.7985 j ( )
Gx ,3 t , e 2
e 2
x d
2 t 4.7985
5
III-C Why Do We Choose the Gaussian Function as a Mask
(1) Among all functions, the Gaussian function has the advantage that the
area in time-frequency distribution is minimal.
( STFT time-domain frequency
domain )
2 2
et /2
e j t dt e f /2
6
Uncertainty Principle (Heisenberg, 1927)
For a signal x(t), if t x t 0 when |t| , then
t f 1/4
2 2 2 2
(t t ) | x (t ) | dt ( f f ) | X ( f ) | df
where t 2f
2
, , ,
| x(t ) | dt | X ( f ) |
2 2
df
f
t | x (t ) | 2
dt f | X ( f ) |2 df
t , .
| x(t ) | | X ( f ) |
2 2
dt df
7
(Proof of Henseinbergs uncertainty principle):
1
2 2 2
t | x (t ) | dt | x (t ) | dt
t f 2
2 2
4 | x(t ) |2 dt | x(t ) |2 dt
2 2
| x (t ) | dt | X ( f ) | df if X(f) = FT[x(t)]
8
2
From Schwarzs inequality x(t ), x(t ) y (t ), y (t ) x(t ), y (t )
d
2
d
2
t
2
| x(t ) | dt | x (t ) | dt tx t x (t )dt tx t x (t )dt / 2
2
2
dt dt
2
d d
tx t x(t ) tx t x (t ) dt / 4 (using |a+b|2 + |ab|2 2|a|2 )
dt dt
2
d 2
t x(t ) x (t ) dt / 4 tx (t ) x (t ) x (t ) x(t ) dt / 4
dt
2
tx(t ) x (t ) t tx(t ) x (t ) t x (t ) x (t ) dt / 4
x (t )
2
dt / 4
1 1
t2 2f t f
16 2 4
9
For Gaussian function
x t e t X f e f
2 2
x t dt e
2 2 t 2
dt ?
2 2
use e ( at bt )
dt / a eb / 4a
t x t dt t e
2 2 2 t 2
2
dt ?
[(m 1) / 2]
2
use t m e a t dt
0 2a ( m1) / 2
1/ 2 n 1 n n
1
, f
4
Gaussian function
1
t f
4
11
Special relation between the Gaussian function and the rectangular
function
3
3
2
2
1
1
f-axis
f-axis
0
0
-1
-1
-2
-2
-3
-3
-4
-4 -4 -3 -2 -1 0 1 2 3 4
-4 -3 -2 -1 0 1 2 3 4
t-axis t-axis
13
x(t) = cos(2 t) when t < 10,
x(t) = cos(6 t) when 10 t < 20,
x(t) = cos(4 t) when t 20
Gabor
5
2
Frequency (Hz)
-1
-2
-3
-4
-5
0 5 10 15 20 25 30
Time (Sec)
14
3 3
2 2
1 1
f-axis
f-axis
0 0
-1 -1
-2 -2
-3 -3
-4 -4
-10 -5 0 5 10 -10 -5 0 5 10
t-axis t-axis
s t exp jt 2 /10 j 3 t for 9 t 1,
s(t) = 0 otherwise,
r t exp jt 2 / 2 j 6t exp (t 4)2 /10
15
1
f-axis
-1
-2
-3
-4
-10 -5 0 5 10
t-axis
16
III-E Properties of Gabor Transforms
Gx t , f e j 2 f e ( t ) x d
2
When k = 0,
When k = 1, Gx t , f e j 2 t f df x t (recovery property)
(b) When x( ) = 1, Gx t , f e j 2 f t f 2
e
(symmetric for the time and frequency domains)
average of Gx t , f e 2 ( t t0 ) average of Gx t0 , f
2 2 2
Gx t , f e
4 ( t ) 2 j 2 f
e x d
4
= 0.2 4
=5
3 3
2 2
1 1
f-axis
f-axis
0 0
-1 -1
-2 -2
-3 -3
-4 -4
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
t-axis t-axis
21
time resolution frequency resolution
(1) Using the generalized Gabor transform with larger
(2) Using other time unit instead of second
t ( sec) f ( Hz)
(1)
(2) Matrix Vector operation
(3)
(4) (
debug)
(5)