Figure 1: An example of the quantization procedure, applied to an analog signal.
Quantization is the process of mapping a continuous range of values into a
finite range of discrete values. A quantizer, usually denoted by , is a system
that:
In the analog case (see the Figure 1), a PAM signal (i.e. a sequence
of analog samples where is an (analog) signal, and is the sampling
period) is transformed into a PCM signal. Therefore, analog quantization
is the process of mapping a continuous range of values (not necessarily
countable) into a finite range of discrete values (necessarily countable) [4].
Notice that, in general, analog signals are 1-dimensional, and that analog
quantization is irreversible.
In the digital case, where is already a PCM signal (a sequence of integers
of finite precision), the output of a quantizer is a sequence of quantization
indexes , and the inverse system, called a dequantizer or inverse quantizer
(denoted by ), can only recovers an approximated version of that it will
be denoted by .
If we define the cardinality operator applied to a digital signal as the number
of different values that such signal can take, i.e., the size of the signal alphabet,
it ususally holds that
As a consequence of this property, the values that , individually, can take will
require less bits to be represented than the values that the original signal
can.
Finally, note that in the case of digital quantization, is only a formal notation
and does not correspond to the reciprocal function of since quantization is
noninvertible [1] (Quantization is not linear transform).
2 Quantization error
Quantization is a lossy process that usually generates a distortion between and . The
quantization error in unpredectible (for this reason is also called quantization noise),
and therefore, we cannot recover after quantization.
We define the quantization error
an the distortion, generally measured as the Mean Squared Error determined by
where denotes the expectation operator.
3 Quantizer design
should be designed to minimize the MSE, and this depends on how the mapping
between the input signal and the quantized signal has been performed.
is defined by a finite set of decision boundaries that describe a set of
cells1
in the signal domain, and a finite set of representation values (see Figure 1), both
with the same cardinality (because there is one representation level per cell). The set
is called the codebook and to their elements, , codewords.
Given a finite number of cells, to minimize the MSE, and are selected
depending on the characteristics of . In general, we need to consider the statistical
distribution of the samples (or vectors of samples, depending on the way we process
the input) in .
Notice that has an impact on the output bit-rate of the quantizer and therefore,
we could be interested in minimizing the RD (Rate/Distortion) tradeoff instead of
simply the MSE. However, such problem in general is addressed by using an entropy
codec at the output of .
4 Scalar quantization
Scalar quantizers map each source sample independently from the other
samples and therefore, a quantization index is produced for each input sample
[2].
5 Vector quantization
In vector quantization, several source samples are quantized simultaneously and a
single index is associated to a vector of source samples. Vector quantization allows to
account for the correlation between source samples directly at the quantizer, which
improves its RD efficiency [1, 3].
6 Uniform quantization
In uniform quantizers, the size of the cells is constant. For example, in a scalar
quantizer, the quantization step size is constant and independent of the input
signal.
Uniform quantizers are used in most A/D (analogic/digital) converters, where it
is expected the generation of uniformely distributed sequences of samples. In the case
of scalar uniform digital quantizers, it is common to have and the input intervals are
of the form .
7 Non-uniform quantization
In order to minimize the MSE, the cells defined by the quantizer cound be of different
size depending on the PDF of . When all the cells do not have the same size we have
designed a non-uniform quantizer.