TM355 – Session 8 – By/ Ayman Metwali – ayman.metwali@gmail.

com

DIGITISATION AND LOSSY COMPRESSION

What is Perceptual Perceptual redundancy ‫ التكرار االدراكي‬refers to the information contained within an audio or visual
redundancy? signal that can be removed without affecting the recipient’s experience of the signal.
‫تشير الى معلومات في االشارات الصوتية او المرئية يمكن حذفها بدون تأثير على مستقبل االشارة‬

BIT RATE AND DISTORTION

Coding the source into the fewest possible number of bits allows
• a lower bit rates
• transmission to be completed faster

Define permissible • The permissible distortion specifies how much the reconstructed version of the source is
distortion? allowed to differ from the original.
.‫يحدد التشوه المسموح به المقدار المسموح بة للنسخة التي أعيد بناؤها من المصدر أن تختلف عن األصل‬

What is the rate rate distortion (RD) characteristic.

distortion • the relationship between the compression level achieved and the resulting distortion
characteristic? ‫العالقة بين مستوى الضغط الذي تم انجازة والتشوة الناتج عن الضغط‬
MCQ: • if our goal is high compression (better coding efficiency) or low bit rates, then we would
expect more distortion (bigger errors) in the reconstructed signal, and vice versa.

SOURCE DIGITISATION
PULSE CODE MODULATION (PCM)
Draw then describe the
Block diagram of a PCM
system?

• The analogue signal sampled to create a pulse amplitude modulation (PAM) signal
• each digital sample then being assigned to one of a finite number of possible discrete
values in a process called quantizing.
• The resulting bitstream goes through further lossless encoding (compression) to minimize
the final transmission bit-rate requirements.

What is the different • Sampling is the process of converting a continuous analogue time signal into a discrete
between sampling and time representation.
quantizing? • Quantizing is the process of assigning every digital sample to one of a finite number of
possible discrete values.

Page 1 of 5
 TM355 – Session 8 – By/ Ayman Metwali – [email protected]

When can you get NO If 𝑓𝑚𝑎𝑥 is the maximum signal bandwidth, there is no loss of information between the
LOSS of information original and sampled signals if and only if the signal is sampled at a rate that is at least
between the original twice 𝑓𝑚𝑎𝑥 →
and sampled signals?
𝑓𝑠 ≥ 2 𝑓𝑚𝑎𝑥
The difference between the original and digital signals is called quantization noise
(quantization error) → quantization is lossy!

S/N ≈ 6.02n dB
n is the number of quantization bits

What will happen if the If the sampling theorem is

sampling theorem is not obeyed ‫اطاع‬, then we
not obeyed? have the aliasing effect.

Define DPCM?
Differential pulse-code modulation (DPCM) is a variant of PCM that also converts a source
analogue signal into a digital representation, but it is able to achieve a lower bit rate by
including sample prediction in its coding.
• Consider the following sequence:
220, 221, 225, 229, 232, 236, 238, 233.
Instead of sending all these values, we could send the receiver the first sample (220),
followed by the differences between successive values.

➔ In this example, the difference values are:

DPCM Samples: 220, +1, +4, +4, +3, +4, +2, −5.

IMAGE AND VIDEO CODING

Define Digital TV(DTV)
Digital TV (DTV)– through compression – allows more channels to be accommodated in a
given bandwidth than is possible in analogue television.

compression algorithms MPEG (Moving Picture Experts Group)

Specify the different MPEG-1:
between MPEG-1, 2, 3, • Mainly used for the efficient storage of moving pictures for multimedia on CD-ROM.
4? • It gave rise to the popular low-resolution video files common interchange format (CIF) and
source intermediate format (SIF), which are still widely used on the internet.
MPEG-2:
• A ‘toolbox’ of optimized compression techniques for DTV systems to support both SD and
HD picture resolutions at bit rates between 4 and 20 Mbit/s.
• MPEG-1 and -2 also include related audio standards– layer 3 – is the basis of MP3 coding.

Page 2 of 5
 TM355 – Session 8 – By/ Ayman Metwali – [email protected]

MPEG-4:
• Originally intended to provide high compression rates, and transmission at bit rates below
64 kbps.
• Adopts ‫ يتبنى‬a different approach to audio via advanced audio coding (AAC).
• The advanced video coding (AVC) standard is now formally incorporated into MPEG-4, and
has become the coding standard for DTV broadcasting and web- streaming services such as
YouTube and BBC iPlayer.
MPEG-7:
• Specifies the way multimedia content can be indexed
)‫• يحدد طريقة فهرسة محتوى الوسائط المتعددة (الفيديو والصوت‬
• searched for in a variety of ways relating to the specific medium
‫• البحث في مجموعة متنوعة من الوسائط عن شي معين‬
• It also covers intellectual property aspects.
‫• كما يغطي جوانب الملكية الفكرية‬
MPEG-21:
• Includes additional digital rights management (DRM) into MPEG systems, enabling the
searching, selecting and managing of rights associated with media objects in multimedia
content.

CODING PRINCIPLES
What is the different Spatial compression (intra-frame compression):
between intera-frame • Exploits ‫ تستغل‬the fact that in many real pictures, considerable similarity (correlation) exists
compression and inter- between neighboring areas of an image, with a correspondingly high level of redundancy in
frame compression? the sampled data.
Important. • This allows each individual picture to be compressed.
• is the basis of the JPEG image compression standard.

Temporal compression (inter-frame compression):

• Exploits the fact that in most sequences, there are very little changes between consecutive
frames.
• This high correlation between consecutive frames offers further significant lossy
compression opportunities, by removing detail without a perceived loss of quality.

What is JPEG Coding? JPEG CODING is A lossy compression standard for color and greyscale images, with an
improvement in coding efficiency of typically between 5 and 15 times over TIFF(lossless)

Page 3 of 5
 TM355 – Session 8 – By/ Ayman Metwali – [email protected]

Draw the four blocks of • JPEG encoding contains four blocks:

JPEG encoding. List
them. • Manage the image pre-processing (the splitting of the source image into
n×n pixel sections known as macroblocks)
• Transform coding of each macroblock using a discrete cosine transform
(DCT)
• Quantization
• Lossless (Huffman) compression.

What is the pre- PRE-PROCESSING

processing?
• The source image is first divided into a number of equally sized
macroblocks, to which the DCT is then applied.
• While a macroblock can be any n × n array of pixels, JPEG generally
uses 8 × 8 pixel macroblocks.
• DCT could be applied to the whole image without breaking it into
macroblocks. However, applying the DCT directly to the whole image
is computationally very intensive.

How does DCT works? DISCRETE COSINE TRANSFORM (DCT)

• DCT converts spatial pixel values into an alternative mathematical
representation.
• in this transform it is easier to see which information is redundant
• Using the transform representation, we can either approximate or
ignore the less significant coefficients (a lossy process), effectively
reducing redundancy.
• DCT makes signal energy compacted into only a few non-zero
coefficients in the transform domain, with the remaining coefficients
being either very small (little energy) or zero (no energy).

What is the Quantization:

quantization in JPEG • is to code the important, higher- valued, low-frequency a.c. coefficients
encoding processing?
with more bits and achieve compression by coding the higher frequency
coefficients with fewer bits.
• Because the eye is not so sensitive to higher frequencies

Page 4 of 5
 TM355 – Session 8 – By/ Ayman Metwali – [email protected]

• High frequencies usually have small coefficient values, so most can be

replaced with a zero value by thresholding
• JPEG quantisation uses Q-tables to determine what level of lossy
compression to apply.
• The coefficient value for each basis function is divided by the corresponding
Q-table entry
• the quantized values are rearranged into the zig-zag scanning pattern
• to achieve further lossless compression we use RLE
• last step is to use Huffman coding.

What is the JPEG

Limitations? JPEG limitations:
• no interactive functionality – JPEG cannot compress a region of interest
(ROI) at a different bit rate from the remainder of the image
‫) بمعدل بت مختلف عن‬ROI( ‫ ضغط منطقة االهتمام‬JPEG ‫ ال يمكن لـ‬- ‫ال توجد وظائف تفاعلية‬ •
‫باقي الصورة‬
• not optimized for either natural images or synthetic computer- generated
images
‫ال يوجد تحسين اذا كانت الصورة طبيعية او تم انتاجها بواسطة الكمبيوتر‬ •
• poor compression of compound documents containing both images and
text
‫سوء الضغط عندما يحتوي المستند على نصوص وصور‬ •
• degraded performance in noisy (wireless/mobile) channel conditions –
quality drops when transmission errors occur.
‫اداءة يقل عند نقل البيانات في القنوات التي بها تشويش‬ •

Page 5 of 5