The Mpeg Handbook Chapter 2' Fundamentals

1
The Mpeg HandbookChapter 2. Fundamentals

• ? ? ?

2
Audio and Video

• What is an audio signal?
• Actual sounds are converted variations in air
  pressure and air velocity to electrical signals
• What is an video signal?
• Television is to allow a moving picture, which is
  a two-dimensional image, which changes as a
  function of time
• This is a three-dimensional information source
• Distance across the screen, distance down the
  screen, and time
• Solution
• Convert the three-dimensional moving image into a
  series of still pictures, taken at the frame rate
• Two-dimensional images are scanned as a series of
  lines

3
Type of video
RGB video
-Single channel television broadcast -PC cable
Component
-Transform-based compression -DVDP
Composite
Chroma
-Differential coding system
4
What is digital signal ? (1)

• Digital equipment is smaller extent than analog
  and at lower cost or same performance at much
  lower cost

5
What is digital signal ? (2)

• Digital signal is used as the input of
  compression system and output of decoding stage
• Quality is determined by ADC and performance of
  the coder

6
Sampling (1)
Hz
Cycle-per-millimeter
Cycle-per-degree
Cycle-per-picture-height Cycle-per-picture-width
7
Sampling (2)
8
Sampling (3)
-Beat-frequency ???? ?? ?? ? ???? ???? ???
?????? ??? ?3? ???
9
Sampling (4)
Fb
Fb
10
Sampling (5)

• Sampling system consist of a pair of filters
• Anti-aliasing filter before the sampling process
• Reconstruction filter after the sampling process
• Low-pass filter

11
Reconstruction (1)

• Input must be band limit by an ideal linear phase
  low-pass filter
• Ideal low-pass filter is a sin(x)/x
• Reconstruction filter has the same frequency
  response

12
Reconstruction (2)

• Optics dont have negative light
• The restriction to positive-only impulse response
  limits the sharpness of optical filters
• Filter must be symmetrical
• non-causal
• Delay of one-half the window period
• (in chapter 3)

13
Aperture effect

• Real sample impulse cannot be infinitely small in
  time
• Zero-Order Hold
• Pulse width equal to the sample period
• About 4dB down at the Nyquist frequency

14
Choice of audio sampling rate (1)

• Nyquist criterion is only the beginning of the
  processing
• For professional products, there is a need to
  operate at variable speed for pitch correction

15
Choice of audio sampling rate (2)

• In the early days of digital audio, video
  recorders were adapted to store audio samples by
  creating a pseudo-video waveform
• Sampling rate is constrained to relate simply to
  the field rate and field structure of TV standard
• 525/60
• 35 blanked lines, 490 lines per frame, 245 lines
  per field for sample
• 60245344.1kHz
• For practical and economic reasons digital audio
  now has essentially three rates to support
• 32 kHz for broadcast, 44.1 kHz for CD, 48 kHz for
  professional use

16
Video sampling structures

• Luminance samples appear at half the spacing of
  colour difference samples
• Other luminance sample is co-sited with a pair of
  colour difference samples
• Figure 2.22 shows 420 subsampling
• Colour data are vertically low-pass filtered

17
The phase-locked loop (1)

• Voltage Controlled Oscillator
• Run at a range of frequencies according to the
  voltage applied to a control terminal
• Driven by a phase error measure between the
  output and some reference

18
The phase-locked loop (2)
19
Quantizing (1)

• Process of expressing some infinitely variable
  quantity by discrete or stepped values

20
Quantizing (2)

• Divides the voltage range up into quantizing
  intervals Q
• Mid-tread quantizer is universally used in audio
  and video
• Audio muting or video blanking is half-way up a
  quantizing interval
• -1/2ltQuantizing errorlt1/2

21
Quantizing error (1)

• Quantizing error waveform can be thought of as an
  unwanted signal
• The quantizing process adds to the perfect
  original
• Additive broadband noise uncorrelated with the
  signal
• Quantizing noise
• At low levels, quantizing error becomes a
  function of the input waveform and the quantizing
  structure
• Unwanted signal becomes a deterministic function
  of the wanted signal
• Distortion rather than noise
• With a large signal , there are so many steps
  involved
• Staircase with many steps appears to be a slope

22
Quantizing error (2)

• The non-linearity of the transfer function
  results in distortion, which produces harmonics
• These harmonics are generated after the
  anti-aliasing filter, and so any which exceed
  half the sampling rate will alias

23
Dither (1)

• Job of decorrelation by making the action of the
  quantizer unpredictable
• Dither causes a slight reduction in the SNR, but
  small price to pay for the elimination of
  non-linearities
• Linearity is an essential requirement for digital
  audio and video
• Ideal quantizer can be dithered by linearly
  adding a controlled level of noise either to the
  input signal or to the reference voltage

24
Dither (2)

• View of system position
• The addition of dither means that successive
  samples effectively find the quantizing intervals
• Quantizing error becomes a function of the dither
• Unacceptable distortion is converted into
  broadband noise
• The dither has resulted in a form of duty cycle
  modulation
• System resolution has been extended indefinitely
• View of transfer function of quantizer position
• Transfer function which is a perfect staircase
  becomes straight

25
Introduction to digital processing

• Only two basic types of element in use
• Combined in some way and supplied with a clock to
  implement
• Strength of binary logic
• Considerable noise and distortion can be
  tolerated
• Signal is compared with a threshold
• Can pass through any number of stages without
  degraded
• Reclocking
• Use of a storage element

26
Logic element (1)

• Exact levels are not of much consequence
• Interfacing between different logic families
• Driving external devices
• There is two states
• High and low
• Positive logic
• High voltage represents a true logic condition
• Low voltage represents a false logic condition
• In binary, the column position specifies the
  power of two
• Parallel system is most convenient inside
  equipment or short distances
• It is inexpensive
• Single signal path is convenient for cables
  between pieces of equipment
• The connectors require fewer pins
• It can be called digital system

27
Logic element (2)

• Important simple gates and their derivatives
• Normally, a high voltage level is a binary 1 and
  a low voltage level is a binary 0

28
Storage elements (1)

• Base memory element in logic circuits is the
  latch
• Edge-triggered device
• Change state at transition
• Level-triggered device
• Change state at level
• Shift register can be made from a series of
  latches
• Connecting the Q output of one latch to the D
  input of the next
• All the clock inputs in parallel
• Useful for converting between serial and parallel
  data formats

29
Storage elements (2)

• When large numbers of bits are to be stored,
  cross-coupled latches are less suitable
• More complicated to fabricate inside integrated
  circuit than dynamic memory
• Consume more current
• In large RAMs, the data bits ard stored as the
  presence or absence of charge in a tiny capacitor
  (MOS)
• Charge will suffer leakage after few milliseconds
• Delay needed is less than this (read out before
  decay)
• Where longer delays are necessary, memories must
  be refreshed periodically
• Desired bit has to be addressed before it can be
  read or written
• Size of the chip package restricts the number of
  pins available
• Large memories use the same address pin
• Bits are arranged internally as rows and columns
• The low address and the column address are
  specified sequentially on the same pins

30
Storage elements (3)

• Basic volatile RAMs will lose data if power is
  interrupted
• Non-volatile RAMs or NVRAMs retain the data in
  the absence of power
• ROM
• A type of memory which is written once is called
  a read-only-memory
• Data are fixed
• UVROM
• Can be written electrically
• Need to be erased by exposure to ultraviolet
  light
• EAROM (electric alterable)
• Can be rewritten electrically a limited number of
  times

31
Binary coding (1)

• Practical digital hardware places a limit on the
  wordlength
• Occur overflow and underflow
• The pulses to be counted are fed to the clock
  input of a D-type latch
• Divide-by-two counter
• As a result of the fixed wordlength, the infinite
  range of real numbers is mapped onto the limited
  range of a binary code of finite wordlength

32
Binary coding (2)

• Mathematically the pure binary mapping from an
  infinite scale to a finite scale is known as
  modulo arithmetic
• For a broadcast standard luminance signal
• Only handle active line
• Sync pulses go off the bottom of the scale
• Small offset in order to handle slightly
  misadjusted inputs
• The extremes of the range are reserved for
  synchronizing
• Colour difference video signals are bipolar
• Blanking is in the centre of the signal range

33
Binary coding (3)

• Digital audio mixing
• Same quantizing interval size, no offset
• achieved by adding sample values
• Others (non-uniform or offset quantizing)
• Cannot be processed
• Binary numbers are not proportional to the signal
  voltage
• If two offset binary streams are added
• May lead to an overflow
• Attenuate by 6.02dB by dividing samples by two
• This approach is not suitable for audio or colour
  difference signals

34
Binary coding (4)

• In twos complement system
• All numbers clockwise from zero are positive and
  have the MSB reset
• All numbers anticlockwise from zero are negative
  and have the MSB set
• MSB is sign bit
• ( 1minus )

35
Binary coding (5)

• Real ADC configure

• MSB inversion may be selectable by an external
  logic level

36
Binary coding (6)

• Linear sum of the two waveforms obtained by
  adding pairs of sample values

37
Binary coding (7)

• Twos complement adding process
• Effectively both twos complement numbers to be
  added contain an offset of half full scale
• Code consists of moving one full rotation round
  the circle of numbers
• Offset has no effect and is effectively
  eliminated
• Sometimes necessary to phase reverse or invert a
  digital signal
• Process of inversion in twos complement
• Invert to form the ones complement, and one is
  added
• Performing a second inversion gives the original
  sample values
• Subtraction can be performed using adding logic
• Radix point
• Numbers to the right of it are added
• 1100.1 is not -4.5, it is -40.5-3.5

38
Binary coding (8)

• When mixing by adding sample values, care has to
  be taken to ensure overflow
• If the MSB of both input is zero, the numbers are
  both positive, thus the sum has the MSB set, the
  output replaced with the maximum positive code

39
Binary coding (9)

• Storage element can be combined with an adder to
  obtain a number of useful functional blocks
• Latch is connected in a feedback loop around an
  adder
• Accumulator
• Discrete time integrator in filtering
• Addition of an inverter allows the difference
  between successive inputs to be obtained
• Digital differentiation

40
Gain control

• When processing digital audio or image data the
  gain of the system will need to be variable
• In digital domain by multiplying each sample
  value by a coefficient
• Multiplication in binary circuits can be
  performed by bit shifting
• The samples to be multiplied must have been
  uniformly quantized

41
Floating-point coding

• Floating-point coding allows a much greater range
  of numbers
• Floating-point is the computers equivalent of
  lossy compression
• Without the inaccuracy of floating-point coding
  by using techniques such as double precision

42
Multiplexing principles

• Multiplexing is used where several signals are to
  be transmitted down the same channel
• In time-division multiplexing the timebase of the
  transmission is broken into equal slots, one for
  each signal

43
Packets

• In multiplexing system, much easier to organize
  if each signal is in the form of data packets of
  constant size
• Each packet consists of two components
• In more complex system, check that packets are
  not lost or repeated
• Packet continuity count in the header

44
Statistical multiplexing

• The multiplexer has to ensure that the total bit
  rate does not exceed the rate of the channel
• With variable-rate inputs, creating null packets
  which are generally called stuffing or packing
• In MPEG environment, statistical multiplexing can
  be extremely useful
• Allow for the varying difficulty of real program
  material

45
Timebase correction (1)

• Accurate control of delay is the essence of
  timebase correction
• Compression result in a variable amount of data
  and effectively the picture period varies
• Buffering will be needed at the encoder and
  decoder

46
Timebase correction (2)

• Shift register approach and the RAM approach to
  delay are very similar
• Addressing of the RAM is given by ring-like
  structure