Scribd
Upload
52 views

Course 18.327 and 1.130 Wavelets and Filter Banks

This document discusses filter banks and the Haar filter bank. It provides the following information: 1) Filter banks can be represented in the time and frequency domains. Perfect reconstruction requires meeting conditions to avoid aliasing and distortion. 2) The Haar filter bank is the simplest non-trivial example of a two-channel filter bank. It is described through its analysis and synthesis filters and matrix representations. 3) Perfect reconstruction means the synthesis bank is the inverse of the analysis bank. This requires the product of the analysis filters to satisfy certain conditions. The document outlines the process for designing perfect reconstruction filter banks.

Uploaded by

djoseph_1
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
52 views

Course 18.327 and 1.130 Wavelets and Filter Banks

This document discusses filter banks and the Haar filter bank. It provides the following information: 1) Filter banks can be represented in the time and frequency domains. Perfect reconstruction requires meeting conditions to avoid aliasing and distortion. 2) The Haar filter bank is the simplest non-trivial example of a two-channel filter bank. It is described through its analysis and synthesis filters and matrix representations. 3) Perfect reconstruction means the synthesis bank is the inverse of the analysis bank. This requires the product of the analysis filters to satisfy certain conditions. The document outlines the process for designing perfect reconstruction filter banks.

Uploaded by

djoseph_1
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

Course 18.327 and 1.

130
Wavelets and Filter Banks

Filter Banks: time domain


(Haar example) and frequency domain;
conditions for alias cancellation
and no distortion

Haar Filter Bank


Simplest (non-trivial) example of a two channel FIR perfect reconstruction filter bank. v0[n] r0[n] t0[n] 2 2 h0[n] f0[n] y0[n] ^ x[n] x[n] Synthesis Analysis t1[n] r1[n] v [n] 2 f1[n] 1 2 h1[n] y1[n]
1 2
-1 0
2

1 2

1 2

1 2

h0[n] =

f0[n] =

1 2

1 2
0 1

h1[n] =

f1[n] =
-1 2

-1 -1 2

Analysis: 1 r0[n] = 2 (x[n] + x[n 1]) y0[n] = r0[2n] 1 y0[n] = (x[2n] + x[2n 1]) 2 Similarly 1 y1[n] = (x[2n] x[2n 1]) 2 lowpass filter downsampler -----------------j

------------------k
3

Matrix form
4 y0[0] y0[1] : : y1[0] y1[1] 4
4 3 1 3 0 1 0 0 1 0 3 1 3
0 3 1 3
: x[-1] x[0] x[1] x[2] : :

1 2

4
3 -1 1 3 0 0 4 0 -1

yo y1

L = B
x

-------------------l

Synthesis t0[n] =

&)')(
&)')(
-

y0[n/2] 0
1 2

n even upsampler n odd

v0[n] =
&)')(
&)')(

( t0[n + 1] + t0[n]) y0[n/2] y0[ n + 1 ]


2

lowpass filter

1 2
1 2

n even n odd

Similarly v1[n] =

1 2
1 2

y1[n/2]
y1[ n + 1 ]
2

n even
n odd

So, the reconstructed signal is ^ x[n] = v0[n] + v1[n]


1 2

(y0[n/2] + y1[n/2])
(y0[ n + 1] - y1[ n + 1 ])
2
2

n even
n odd

1 2

i.e. ^
x[2n-1] =
^ x[2n]
^ So x[n] =
1 2
1
2

(y0[n] y1[n]) = x[2n-1]


from j and k
(y0[n] + y1[n]) = x[2n]
x[n] Perfect reconstruction!

In general, we will make all filters causal, so we will have ^


x[n] = x[n n0] PR with delay

Matrix form 4 ^ x[-1] ^ x[0] 1 = ^ 2 x[1] ^ x[2]


4

4 4 1 0 1 0 3
3 3

4 4 -1 0 1 1

0 1 0 1 4 4

0 -1 -1 1 4 4

4 y0[0] y0[1] 4 4 y1[0] y1[1] 4

^ x

y0

LT

BT
y1

----------------m

Perfect reconstruction means that the synthesis bank is the inverse of the analysis bank.
^ = x x

LT

BT
&'(

= I
B

&'(
W-1

W Wavelet transform matrix In the Haar example, we have the special case W1 = WT orthogonal matrix So we have an orthogonal filter bank, where Synthesis bank = Transpose of Analysis bank f0[n] = h0[- n] = h1[- n] f1[n]

Perfect Reconstruction Filter Banks General two-channel filter bank


H0(z)
x[n]
r0[n]

y0[n]

3 3

t0[n]

v0[n]

F0(z)

H1(z)

r1[n]

y1[n]

t1[n]

F1(z)

^ x[n] v1[n]

z-transform definition: X(z) = x[n]z-n


n=-

Put z = ei w to get DTFT

10

Perfect reconstruction requirement: ^ x[n] = x[n - ?] ^ X(z) = z- ? X(z) (? time delays)

H0(z) and H1(z) are normally lowpass and highpass, but not ideal H1(w) 1 H1(w) H0(w)

p w 0 -p p 2 2 Downsampling operation in each channel can produce aliasing

-p

11

Lets see why: Lowpass channel has Y0(z) = {R0(z) + R0(-z)} (downsampling)

= {H0(z) X (z) + H0(-z)X(-z)} In frequency domain: X(z) X(w) X(-z) X(w + p) w X(z) X( 2 )
Y0(w) = {H0(
w ) 2

or X(eiw)

X(

w ) 2

+ H0(

w 2

+ p)X(

w 2

+ p)}
12

Suppose X(w) = 1 (input has all frequencies) Then R0(w) = H0(w), so that after downsampling we have
R0(
w 2

+ p)

R0( w )
2

R0(

w 2

+ p)

Y0(w) =
-p 0 p
aliasing

Goal is to design F0(z) and F1(z) so that the overall system is just a simple delay - with no aliasing term: V0(z) + V1(z) = z- ? X(z)
13

V0(z) = F0(z) T0(z) = F0(z) Y0(z2) (upsampling)

= F0(z){ H0(z) X(z) + H0(-z) X(-z)} V1(z) = F1(z){ H1(z) X(z) + H1(-z) X(-z)}

So we want
{F0(z) H0(z) + F1(z) H1(z) } X(z)
+ {F0(z) H0(-z) + F1(z) H1(-z) } X(-z)
14

= z- ? X(z)

Compare terms in X(z) and X(-z): 1) Condition for no distortion (terms in X (z) amount to a delay) F0(z) H0(z) + F1(z) H1(z) = 2z- ?
--------------j

2) Condition for alias cancellation (no term in X(-z)) F0(z) H0(-z) + F1(z) H1(-z) = 0
--------------k

To satisfy alias cancellation condition, choose F0(z) = H1(-z) F1(z) = -H0(-z)

----------------------l
15

What happens in the time domain? F0(z) = H1(-z) F0(w) = H1(w + p) = h1[n] (-z)-n
n

= (-1)n h1[n] z-n


n

So the filter coefficients are f0[n] = (-1)n h1[n] f1[n] = (-1)n+1 h0[n] Example h0[n] = { a0, a1, a2}
h1[n] = { b0, b1, b2}

alternating signs rule


f0[n] = { b0, -b1, b2 } f1[n] = {-a0, a1, -a2}
16

Product Filter Define P0(z) = F0(z) H0(z)

------------------------------m

Substitute F1(z) = -H0(-z) , H1(z) = F0(-z) in the zero distortion condition (Equation j) F0(z) H0(z) - F0(-z) H0(-z) = 2z- ? i.e. P0(z) - P0(-z) = 2z-? ----------------------------n

Note: ? must be odd since LHS is an odd function.

17

Normalized Product Filter Define P(z) = z? P0(z)

----------------------------o

P(-z) = -z? P0(-z) since ? is odd So we can rewrite Equation n as z- ? P(z) + z- ? P(-z) = 2z- ? i.e. P(z) + P(-z) = 2
---------------------------p

This is the condition on the normalized product filter for Perfect Reconstruction.
18

Design Process 1. Design P(z) to satisfy Equation p. This gives P0(z). Note: P(z) is designed to be lowpass. 2. Factor P0(z) into F0(z) H0(z). Use Equations l to find H1(z) and F1(z). Note: Equation p requires all even powers of z (except z0) to be zero: p[n]z-n + p[n](-z)-n = 2
n n

&'(

p[n] =

1 ; 0 ;

n=0 all even n (n 0)

19

For odd n, p[n] and p[n] cancel. The odd coefficients, p[n], are free to be designed according to additional criteria. Example: Haar filter bank H0(z) = 2 (1 + z-1) F0(z) = H1(-z) =
1

H1(z) = 2 (1 z-1)

1 (1 + z-1) 2 1 F1(z) = -H0(-z) = 2 (1- z-1) 1 P0(z) = F0(z) H0(z) = 2 (1 + z-1)2

20

10

So the Perfect Reconstruction requirement is P0(z) P0(-z) =


1 1 + 2z-1 + z-2) 2
1 1 2z-1 + z-2) 2

= 2z-1 P(z) = z? P0(z) =


lm

? = 1

1 -1 2 (1 + z)(1 + z )

z Zeros of P(z): 1+z = 0 1 + z-1 = 0 1 Re

2nd order zero at z = -1

21

11

You might also like