Beruflich Dokumente
Kultur Dokumente
QUANTIZATION
Input
visual
information Source Channel
A/D Modulation
encoder encoder
Channel
Received
visual
information Source Channel
D/A decoder decoder Demodulation
Receiver
Input
visual
information
Source
A/D
encoder
Retrieved
visual
information
Source
D/A decoder
Retrieval
Input Codeword
information string
Codework
Transformation Quantization
assignment
Codeword Reconstructed
string information
Codeword Inverse
decoder transformation
♦ Simplest
♦ Most popular
♦ Conceptually, of great importance.
2.2.1 Basics
2.2.1.1 Definitions
y y9
4
y8
3
y7
2
y6
1
y5 x
-3.5 -2.5 -1.5 -0.5 0 0.5 1.5 2.5 3.5
d 1 =-∞ d2 d3 d4 d5 d6 d7 d8 d9 d 10 =∞
y4
-1
y3
-2
y2
-3
y1
-4
y8
3.5
y7
2.5
y6
1.5
y5
0.5 x
-4.0 -3.0 -2.0 -1.0 0 1.0 2.0 3.0 4.0
y4
d 1 =-∞ d2 d3 d4 -0.5 d6 d7 d8 d 9 =∞
d5
y3
-1.5
y2
-2.5
y1
-3.5
d
N i+1
MSEq = ∑ ∫ ( x − Q( x))2 f ( x)dx (2. 3)
i=1 d X
i
f x (x) : probability density function (pdf)
. y
granular overload
quantization quantization
noise noise
0.5
x
-4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5 1 1.5 2.0 2.5 3.0 3.5 4.0 4.5
-0.5
overload
quantization
noise
y
y9
4
y8
3
y7
2
y6
1
y5
x
-4.5 -3.5 -2.5 -1.5 -0.5 0 0.5 1.5 2.5 3.5 4.5
d1 d2 d3 d4 d5 d6 d7 d8 d9 d 10
y4
-1
y3
-2
y2
-3
y1
-4
Granular Zero
quantization overload
noise quantization
noise
0.5
x
-4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5 1 1.5 2.0 2.5 3.0 3.5 4.0 4.5
-0.5
Zero
overload
quantization
noise
σ x2
SNRms = 10 log = 10 log N 2 . (2. 5)
10 σ 2 10
q
♦ Zero mean:
In Table 2.1, all distributions: a zero mean.
If the mean is not zero, only a shift in input is
needed when applying these results.
♦ Unite variance:
In Table 2.1, all distributions: a unit variance.
If the standard deviation is not unit, the
tabulated step size needs to be multiplied by
the standard deviation.
1.0
0.9
Error ratio
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
2 6 10 14 18 22 26 30 34 38
2.1519
1.3439
0.7560
0.2451
X
-∞ -1.7479 -1.0500 -0.5005 0.5005 1.0500 1.7479 ∞
Statistical parameters
Interval
Input x index Quantization Output y
Statistical Quantization
Buffering decoder
analysis encoder
Reconstruction
level
transmitter receiver
Input x Output y
Quantization Quantization
encoder decoder
Buffering Buffering
Statistical Statistical
analysis analysis
transmitter receiver
Q1
Q2
input x
output y
Q3
Q4
Statistical
Buffering QL
analysis
2.5 PCM
f(t)
t
0
PAM
t
0
PWM
t
0
PPM
t
0
d1
0000 y1
d2 10 11
12
0001 y2 13
9 14
d3
0010 y3 8 15
d4
7
0011 y4 16
d5 6
0100 y5
d6 1 2 3 4 5
0101 y6 17
d7
0110 y7
d8
0111 y8 18 x
d9
1000 y9
d 10
1001 y 10
d 11 19
1010 y 11
d 12
1011 y 12
d 13 20
1100 y 13
d 14
1101 y 14 21
d 15
1110 y 15 22
29
d 16
23 28
1111 y 16 24 27
d 17 25 26
0101 0101 0101 0101 0101 0100 0011 0011 0010 0001 0001 0001 0010 0010 0011
0100 0101 1000 1010 1100 1101 1110 1110 1111 1111 1111 1111 1110 1110
[jayant 1970] N. S. Jayant, ‘‘Adaptive delta modulation with one- bit memory,’’
Bell System Technical Journal, 49, 321- 342, March 1970.
[jayant 1973] N. S. Jayant, ‘‘Adaptive quantization with one word memory,’’Bell System
Technical Journal, 52, 1119-1144, September 1973.
[li 1995] W. Li and Y.-Q. Zhang, ‘‘Vector- based signal processing and
qunatization for image and video compression,’’Proceedings of the IE EE, vol.
83, no. 2, pp. 317-335, February 1995.
[paez 1972] M. D. Paez and T. H. Glisson, ‘‘Minimum mean squared error qunatization
in speech PCM and DPCM Systems,’’IEEE Trans. on Communications, pp. 225-230,
Indexes 35
April 1972.