8/19/2017

Introduction

Source Coding and Compression

Introduction

Source Channel Modu-

Encoder Encoder lator
Capture

Channel

Display
Loudspeaker

Source Channel Dem odu-

Decoder Decoder lator

o o
Heiko Schwarz Source Coding and Compression 1 / 30 Heiko Schwarz Source Coding and Compression 3 / 30

Introduction Motivation for Source Coding Introduction Motivation for Source Coding

Motivation for Source Coding Practical Source Coding Problems

File compression (text file, office document, program code, ...)

Example: Example 80 MByte down to 20 MByte (20%)
Source coding or compression is required for efficient transmission or
storage, leading to one or both of the following benefits: Audio compression
Transmit more data given throughput (channel capacity or storage space) Stereo with sampling frequency of 44.1 kHz
Use less throughput given data Each sample being represented with 16 bits
= Raw data rate: 44.1162 = 1.41 Mbit/s
= Typical data rate after compression: 64 kbit/s (4.5%)
Typically, source coding or compression are considered enabling technologies,
i.e., technologies that enable an application Image compression
Original picture size: 30002000 samples (6 MegaPixel)
Examples for source coding applications: 3 color components (red, green, blue) and 1 byte (8 bit) per sample
gzip, compress, winzip, ... = Raw file size: 300020003 = 18 MByte
= Typical compressed file size: 1 MByte (5.6%)
Mobile voice, audio, and video transmission
Internet voice, audio, and video transmission Video compression
Digital television Picture size of 19201080 pixels and frame rate of 50 Hz
MP3 and portable video players (iPod, ...) Each sample being digitized with 8 bit
Digital Versatile Discs (DVDs) and Blu-Ray Discs 3 color components (red, green, blue)
= Raw data rate: 192010808503 = 2.49 Gbit/s
= Typical compressed data rate: 12 Mbit/s (0.5%)
o o
Source Coding and Compression 4 / 30 Heiko Schwarz Source Coding and Compression 5 / 30

1
 8/19/2017
Introduction Motivation for Source Coding Introduction Analog-to-Digital Conversion

Source Coding in Practice Pulse-Code Modulation

Source coding often enables applications:

Digital television (DVB-T)
Internet video streaming (YouTube)

Source coding makes applications economically feasible

Distribution of digital images
High definition television (HDTV) over IPTV

Many applications use source coding techniques

Software is often distributed in compressed form
Audio data are typically compressed (MP3, AAC)
Mobile audio players (IPod,...) and mobile phones
Audio download (ITunes) and streaming services (Internet radio)
Digital images are typically compressed (JPEG)
Compression is often done in camera
Picture found on web sites are compressed
Digital video data are typically compressed (MPEG-2, H.264/AVC)
Output of video cameras, optical discs
Video streaming (Youtube, Internet TV)
About 70% of the bits in the Internet are compressed video data
o o
Source Coding and Compression 6 / 30 Heiko Schwarz Source Coding and Compression 7 / 30

Introduction Analog-to-Digital Conversion

Analog-to-Digital Conversion: Overview

o
Heiko Schwarz Source Coding and Compression 8 / 30

2
 8/19/2017
Introduction Analog-to-Digital Conversion Introduction Analog-to-Digital Conversion

Why Analog-to-Digital Conversion? One-Dimensional Signal Example

o
Source Coding and Compression 10 / 30

Introduction Analog-to-Digital Conversion Introduction Analog-to-Digital Conversion

Two-Dimensional Signal Example Two-Dimensional Signal Example

Pictures are two-dimensional spatial signals

Videos are three-dimensional spatio-temporal signals Below quantization of picture Lena with different bits/sample
Below sampling of picture Lena with different spatial sampling rates k = 1, 2, 4, and 8 bits/sample (from left to right)
8 x 8, 16 x 16, 32 x 32, and 128 x 128 samples (from left to right) Each The spatial sampling rate is fixed to 128x128
sample is represented with n = 8 bits
Each square represents average of luminance values it covers

o o
Heiko Schwarz Source Coding and Compression September 21, 2013 12 / 30 Heiko Schwarz Source Coding and Compression September 21, 2013 13 / 30

3
 8/19/2017
Introduction Communication Problem Introduction Communication Problem

Basic Communication Problem Example: JPEG (1:10 Compression)

o
Heiko Schwarz Source Coding and Compression September 21, 2013 16 / 30

Introduction Communication Problem Introduction Communication Problem

Example: JPEG (1:50 Compression) Example: H.265/HEVC (1:50 Compression)

o o
Source Coding and Compression 17 / 30 Heiko Schwarz Source Coding and Compression September 21, 2013 18 / 30

4
 8/19/2017
Introduction Communication Problem Introduction Communication Problem

Geometrical Interpretation Transmission System

Raw data Source Channel Modu-

Capture Encoder Encoder lator
1:10
Channel

1:50 Display
Loudspeaker

Source Channel Demodu-

1:250 (typical for video) Decoder Decoder lator

o o
Source Coding and Compression 19 / 30 Heiko Schwarz Source Coding and Compression 20 / 30

Introduction Communication Problem Introduction Communication Problem

Practical Communication Problem Scope of This Course

Source codecs are primarily characterized in terms of:

Throughput of the channel, a characteristic influenced by
transmission channel bit rate and
amount of protocol and error-correction coding overhead incurred by
transmission system
Distortion of the decoded signal, which is primarily induced by
source encoder and
by channel errors introduced in path to source decoder

o o
Source Coding and Compression 21 / 30 Heiko Schwarz Source Coding and Compression 23 / 30

5
 8/19/2017
Introduction Communication Problem Introduction Communication Problem

Transmission Channels and Optical Storage Media Types of Compression

Fixed transmission lines:

ISDN line: 64 kbit/s Lossless coding:
ADSL: 6 Mbit/s Uses redundancy reduction as the only principle and is therefore reversible
VDSL: 25 Mbit/s or 50 Mbit/s Also referred to as noiseless or invertible coding or data compaction
Mobile networks: Well known use for this type of compression for data is Lempel-Ziv coding
(gzip) and for picture and video signals JPEG-LS is well known
GSM: 15 kbit/s
EDGE: 474 kbit/s (max)
HSDPA: 7.2 Mbit/s (peak) Lossy coding:
LTE: 300 Mbit/s (peak) Uses redundancy reduction and irrelevancy reduction and is therefore not
reversible
Broadcast channels It is the primary coding type in compression for speech, audio, picture, and
DVB-T: 13 Mbit/s (16QAM) video signals
DVB-S: 38 Mbit/s (QPSK)
DVB-C: 38 Mbit/s (64QAM) The practically relevant bit rate reduction that is achievable through lossy
compression is typically more than an order of magnitude larger than with
Optical storage media lossless compression
Compact Disc (CD): 650 MByte with 1.41 Mbit/s (12 cm) Well known examples are for audio coding are the MPEG-1 Layer 3 (mp3), for
Digital Versatile Dics (DVD): 4.7 GByte with 10.5 Mbit/s (DVD-5-SS-SL) still picture coding JPEG, and for video coding H.264/AVC
Blu-Ray Disc (BRD): 50 GByte with 36 Mbit/s (12 cm, DS-DL)
o o
Source Coding and Compression 24 / 30 Source Coding and Compression 25 / 30

Introduction Distortion/Quality Measures

Distortion Measures

The use of lossy compression requires the ability to measure distortion

Often, the distortion that a human perceives in coded content is a very
difficult quantity to measure, as the characteristics of human perception are
complex
Perceptual models are far more advanced for speech and audio codecs than
for picture or video codecs
In speech and audio coding,
Perceptual models are heavily used to guide encoding decisions
Listening tests are used to determine subjective quality of coding results
In picture and video coding,
Perceptual models have limited use to guide encoding decisions (mainly
focusing on properties of the human visual system)
Viewing tests are used to determine subjective quality of coding results
This lecture: Use of objective distortion measures such as MSE and SNR

o
Source Coding and Compression 26 / 30

6
 8/19/2017
:Rate Distortion Theory:
In lossy coding, the reconstructed signal is not identical to the source
signal, but represents only an approximation of it. A measure of the
deviation between the approximation and the original signal is
referred to as distortion. Rate distortion theory addresses the
problem of determining
the minimum average number of bits per sample that is required for
representing a given source without exceeding a given distortion.

The greatest lower bound for the average number of bits is referred
to as the rate distortion function and represents a fundamental
bound on the performance of lossy source coding algorithms,
similarly as the entropy rate represents a fundamental bound for
lossless source coding. For deriving the results of rate distortion
theory, no particular coding technique is assumed. The applicability
of rate distortion theory includes discrete and continuous random
processes.

Next

Move to

Lossless Coding Techniques