Non-Uniform Scalar Quantization
and vector Quantization
EEE/ETE 422
Saif Ahmed (SfA)
Recall: Uniform Quantization
Types of Quantization
• There are two types of Quantization - Uniform
Quantization and Non-uniform Quantization.
• The type of quantization in which the
quantization levels are uniformly spaced is
termed as a Uniform Quantization. The type of
quantization in which the quantization levels are
unequal and mostly the relation between them is
logarithmic, is termed as a Non-uniform
Quantization.
Uniform Quantization
• There are two types of uniform quantization. They are Mid-
Rise type and Mid-Tread type. The following figures
represent the two types of uniform quantization.
Uniform Quantization
• The Mid-Rise type is so called because the origin lies in the middle of a raising part of
the stair-case like graph. The quantization levels in this type are even in number.
• The Mid-tread type is so called because the origin lies in the middle of a tread of the
stair-case like graph. The quantization levels in this type are odd in number.
• Both the mid-rise and mid-tread type of uniform quantizers are symmetric about the
origin.
Non Uniform Quantization
• Although uniform quantization is straight forward and appears to be a
natural approach it may not be optimal.
• Suppose is much more likely to be in one region than in others. It is
reasonable to assign more reconstruction levels to that region.
• If falls rarely between f and f, the reconstruction level is rarely used.
Important note
• Uniform quantization is an ideal choice if
the probability distribution of the given
symbols are equal.
• With a non uniform probability distribution
function we use non uniform quantization.
A hybrid approach
• Rearranging reconstruction levels so that they all lie between and makes
more sense. Quantizers in which reconstruction and transition levels do
not have even spacing is called non-uniform quantization.
• Quantize g with a uniform quantization and then perform the inverse
nonlinearity.
Vector Quantization
A 2D quantizer operates on 2 source samples u = (u1, u2)
at a time; i.e., the source alphabet is U = R2. Assuming the
source symbols, the joint pdf is then f(u)= fU (u1)fU (u2)
Quantization Error
• For any system, during its
functioning, there is always a
difference in the values of its input
and output. The processing of the
system results in an error, which is
the difference of those values.
• The difference between an input
value and its quantized value is
called a Quantization Error.
A Quantizer is a logarithmic
function that performs
Quantization rounding off the
value. An analog-to-digital
converter (ADC) works as a
quantizer.
Companding
• In telecommunication and signal processing, companding is a
method of mitigating the detrimental effects of a channel with
limited dynamic range. The name is a portmanteau of the words
compressing and expanding, which are the functions of a
compander at the transmitting and receiving end respectively.
Comparison of µ-Law and A-Law
A-law Companding Technique
• Uniform quantization is achieved
at A = 1 where the characteristic
curve is linear and no
compression is done.
• A-law has mid-rise at the origin.
Hence, it contains a non-zero
value.
• A-law companding is used for
PCM telephone systems
µ-law Companding Technique
• Uniform quantization is
achieved at µ = 0, where the
characteristic curve is linear and
no compression is done.
• µ-law has mid-tread at the
origin. Hence, it contains a zero
value.
• µ-law companding is used for
speech and music signals.
