Presentation is loading. Please wait.

Presentation is loading. Please wait.

University of Canberra Advanced Communications Topics

Similar presentations


Presentation on theme: "University of Canberra Advanced Communications Topics"— Presentation transcript:

1 University of Canberra Advanced Communications Topics
Television Broadcasting into the Digital Era Lecture 3 Audio/Video Compression MPEG-2 Multiplexing by: Neil Pickford 1

2 Enabling Technologies
Source digitisation (Rec 601 digital studio) Compression technology (MPEG, AC-3) Data multiplexing (MPEG) Display technology (large wide screens) Transmission technology (modulation) Production Digital TV has Key Technologies that make it possible. Most production within the current TV stations already happens in the digital domain using standards such as Rec 601 digital video. It only becomes analog when it is transmitted over the air to the viewer. Display technology has not reached the level needed for HDTV to be fully implementable at present.

3 Video Bitrate - HDTV = 1.24416 G bits / sec for Interlace Scan or
2 M pixels * 25 pictures * 3 colours * 8 bits = G bits / sec for Interlace Scan or = G bits / sec for Progressive We need to Compress this a bit!

4 Compression Technology
When low bandwidth analog information is digitised the result is high amounts of digital information. 5 MHz bandwidth analog TV picture º Mb/s digital data stream. 270 Mb/s would require a bandwidth of at least 140 MHz to transport Compression of the information is required Compression technology is also an enabling thechnology. When you digitise video you end up with massive amounts of real time data. In the above example you would need 20 channels to transmit one digital video signal. We have to compress the television signal so there is less data, allowing it to fit in a normal channel

5 Compression - Types Two types of compression available
Loss-less compression 2 to 5 times Lossy compression 5 to 250 times Loss-Less compression produces exactly the same data out as went into the process, like over a telephone modem. Lossy Compression allows changes to occur in the data such that subtle approximations are made to the images or sounds that the viewer will not be able to notice.

6 Compression - Loss-less Types
Picture differences - temporal Run length data coding - GIF = 1 + 4x 21 bits source = 12 bits compressed Huffman coding - PKZIP Short codes for common blocks Longer codes for uncommon blocks Lookup tables Temporal Picture Differences rely on the premise that most of the background in pictures does not change from picture to picture, so why transmit it more than once. A difference is calculated by subtraction and only the information that has changed is transmitted. Run Length encoding exploits repeating sequences or data patterns

7 Compression - Lossy Types
Quantisation - rounding Motion vectors Prediction & interpolation Fractal coding Discrete cosine transform (DCT) Quantisation - A 8 or 16 bit signal may not need that level of resolution, 4 or 6 bits may suffice. It might, however make things a bit more fuzzy. Motion Vectors - is a technique where common pixel blocks are identified from picture to picture and their movement transmitted, instead or retransmitting all the information for the blocks. Eg a hand moving, the pixel blocks displaying the hand are identified and the information transmitted that it they moved X pixels in direction Y. The main problem with this technique is the high level of processor power needed to carry out a pixel block search and match. Prediction & Interpolation use averaging to determine data between known points without having to transmit it. DCT is used by MPEG-2 along with Differencing, Motion Vectors, Prediction and Interpolation.

8 Compression - DCT Here is an example of DCT compression. A simple 8x8 pixel area around the “1” on the calendar has been compressed using the DCT. The original and compressed data values in an 8x8 matrix are shown. Notice the 64 original values have been reduced in this particular example to 4 non zero numbers. The numbers in the DCT matrix represent a frequency distribution of the H & V pixel information in the original picture. The quantisation level of the final numbers can also be reduced with typically half of the matrix being zero or very close to zero not requiring transmission. The reverse process produces a pixel block which is a very close approximation to the original, even when some of the elements have been quantised. 8x8 Pixels

9 MPEG-2 - I, P & B Frames Uncompressed SDTV Digital Video Stream Mb/s Picture 830kBytes Picture 830kBytes Picture 830kBytes Picture 830kBytes I Frame B Frame B Frame P Frame 100 kBytes 12 kBytes 12 kBytes 33 kBytes MPEG-2 Compressed SDTV Digital Video Stream Mb/s I - intra picture coded without reference to other pictures. Compressed using spatial redundancy only P - predictive picture coded using motion compensated prediction from past I or P frames B - bidirectionally-predictive picture using both past and future I or P frames MPEG has different types of frames which allow interpolation and prediction to be used to reduce the amount of data that needs to be sent. Three types of frames I, P & B frames. I frames have around 9 times less data, P frames 25 times less and B frames 70 times less data than the original frame. In this example the total compression is 43 times. These frames are usually sequenced in a 12 frame Group Of Pictures (GOP) sequence. Typically structured I B B P B B P B B P B B I B B P B B P B B P B B I etc

10 MPEG-2 - Formats ML & HL MPEG-2 defines profiles & levels
They describe sets of compression tools DTTB uses main profile. Choice of levels Higher levels include lower levels Level resolution Low level (LL) by 288 SIF Main level (ML) by 576 SDTV High level (HL) 1920 by 1152 HDTV Profiles and Levels within MPEG-2 define sets of tools or syntaxes for compressing the picture information. If you have a higher level toolbox than you can handle the lower level compression modes, however if you have a middle level tool set you are unable to process images built with the high level tool set. For SDTV you need Main Level (ML) where as for HDTV you need High Level (HL). If you want to process HDTV you must have a HL decoder. During HL only transmissions a ML decoder will stop decoding and go “Black”. Although HDTV may not be ready at the start of the Digital TV era, the decoders must have a HL decoder otherwise when HDTV is available those decoders will not work. A more inefficient solution which may need to be adopted in Britain, which is installing ML only decoders, is to always transmit a ML signal along with the HDTV. Unfortunately this reduces the data rate available to HDTV by 3-4 Mb/s.

11 MPEG Profiles and Levels
MAX. BIT-RATE 300 Mbit/s 100 Mbit/s 80 Mbit/s 60 Mbit/s 40 Mbit/s This diagram is a summary of the profile and levels of MPEG. “Profile” is what has been used in MPEG to mean a subset of the syntax, for example if you look at the main profile, which is the one most commonly used, the difference between it and the simple profile, is that bidirectionally coded frames can be used. Similarly as we go up the scale of profiles, extra functionality is added. In almost every case, but not in the 4:2:2 case, it’s an onion ring where each one further down the scale is a simple sub-set of the ones further up the scale. “Levels” is a term that is used for constraints on parameters. Typical parameters that this covers are Video resolution or Bitrate. The diagram is attempting to highlight the bitrate by the height of the cylinders. If we look at the MPEG Main Profile at Main Level the maximum allowed bitrate for that is 15 Mbits/s, the maximum allowed video resolution corresponds to Recommendation 601 in it’s 25 & 50 Hz flavors. The next one in yellow is Main Profile at High Level, this is what has been chosen for High Definition TV. In addition to having the enhanced resolution, you are allowed a higher bit rate which is 80 Mbits/s. If we look at the 4:2:2 Profile (Green). In addition to being able to have 4:2:2 encoding as well as the 4:2:0 (In other words having full horizontal resolution relative to Rec 601) you are also allowed to have a much higher bit rate. The reason for the much higher bit rate is, in addition, in a contribution application wanting to have the highest possible quality prior to transmission, you may also want to have more frequent occurrences of Intra-coded frames which are the points at which you can make editing. This results in less efficient coding, so therefor you need a higher bit rate to maintain the quality. 20 Mbit/s HIGH HIGH-1440 4:2:2 LEVELS MAIN HIGH SPATIALLY SCALABLE SNR SCALABLE LOW PROFILES MAIN SIMPLE 4

12 MP@ML MP@HL All decoders sold in Australia will
be capable allowing all viewers access to HD resolution when it becomes available

13 Digital Audio - Multichannel
Two sound coding systems exist for Digital TV MPEG 1 & 2 Dolby AC-3 Cover a wide variety of Audio Applications DVB VCD and S-VCD DAB, DBS, DVD Cinema (Film) Computer Operating Systems (Windows) Professional (ISDN codecs, tapeless studio, ….)

14 Multichannel Sound TV C LFE R L Ls Rs

15 Masking Both use perceptual audio coding that exploits a psychoacoustic effect known as masking

16 Multichannel Sound - MPEG 1/2
MPEG Audio Layer II was developed in conjunction with the European DVB technology Uses Musicam Compression with 32 sub bands MPEG 1 is basic Stereo 2 channel mode MPEG 2 adds enhancement information to allow 5.1 or 7.1 channels with full backwards compatibility with the simple MPEG 1 decoders MPEG 1 is compatible with Pro-Logic processing. Bitrate 224 kb/s MPEG 1 Bitrate kb/s MPEG 2 5.1

17 MPEG Audio Encoder Audio Bit Stream 32 Subbands O/P Subband Filter
Quantiser & Coder Frame Packer Audio In 2 x kb/s 2 x 768 kb/s Bit Allocation Coding of Side Information Psycho- Acoustic Model

18 MPEG Audio Decoder Audio Bit Stream Inverse Subband Filter
Frame Unpacker De-Quantiser Inverse Subband Filter Audio Out 2 x kb/s 2 x 768 kb/s Decoding of Side Information

19 Multichannel Sound - Dolby AC-3
Dolby AC-3 was developed as a 5.1 channel surround sound system from the beginning. Compression Filter bank is 8 x greater than MPEG 2 (256) Must always send full 5.1 channel mix One bitstream serves everyone Decoder provides downmix for Mono, Stereo or Pro-Logic Listener controls the dynamic range, Audio is sent clean Bitrate 384 kb/s or 448 kb/s Dialogue level passed in bit-stream

20 AC-3 Coding Block Diagram

21 AC-3 Multichannel Coder
5.1-ch Encoder 5.1-ch Decoder C LS LS RS RS LFE LFE Encoder Decoder

22 AC-3 Stereo Decoder 5.1-ch Decoder 5.1-ch Encoder Matrix Encoder
L L R R Lo C 5.1-ch Decoder 5.1-ch Encoder C Matrix Ro LS LS RS RS LFE LFE Encoder 2-channel Decoder

23 Problem of AC-3 An AC-3 stereo decoder is more complex than an AC-3 multichannel decoder With AC-3, end users with stereo receivers must pay extra so that others can enjoy multichannel sound

24 MPEG-2 Multichannel Coder concept
MPEG-2 Encoder MPEG-2 Decoder Down mix R L C LS RS LFE Ro Lo MPEG-1 Encoder MPEG-1 Decoder Re matrix R L C LS RS LFE Extension Encoder Extension Decoder

25 Low cost 2-channel decoder
MPEG-1 Encoder MPEG-1 Decoder Lo Lo L Ro Ro R Down mix C LS RS T2 Extension Encoder T3 T4 LFE LFE 2-channel Decoder MPEG-2 Encoder  Low cost 2-channel decoder

26 Compatibility with Pro Logic
S Surround Encoder MPEG-1 Encoder MPEG-1 Decoder R L C S Pro Logic Decoder Rt Lt Ro Lo

27 Compatibility with Pro Logic (2)
LS RS LFE T2 T3 T4 MPEG-1 Encoder Extension Encoder Down mix MPEG-2 Encoder MPEG-1 Decoder Pro Logic Decoder R L C S Rt Lt Ro Lo

28 Widely Available All major MPEG-2 Video decoders incorporate 2-channel or 5.1 channel MPEG-2 Audio Several dedicated MPEG-2 multichannel decoders More than 100 Million decoders world-wide

29 Studio Multichannel Sound
Present AES3 PCM Audio does not cater for 5.1 channel surround. Dolby has produced a system called Dolby E Handles 6-8 audio inputs Uses low compression 3-4:1 Can be transported/stored on 2ch PCM audio equipment Incorporates time stamps and is segmented at the video frame rate allowing editing on video frame boundaries

30 Digital Television - Types
Satellite (DBS) DVB-S Program interchange Direct view / pay TV SMATV Downlink Uplink What are the type of Digital Television? Satellite uses a central uplink to provide regional or national coverage. You need to have a satellite dish and down converter. Because accuracy in pointing is required this is a fixed service.

31 Digital Television - Types
Cable HFC - pay TV MATV DVB-C / 16-VSB Fibre Main Coax Tethered cable systems coaxial or fibre. You need to be connected to a permanent cable. Again this is a relatively fixed system. Spur Tee Tap

32 Digital Television - Types
Terrestrial (DTTB) DVB-T / 8-VSB Free to air TV (broadcasting) Narrowcasting/value added services Untethered - portable reception DTTB allows both fixed or portable operation.

33 Enabling Technologies
Source digitisation (Rec 601 digital studio) Compression technology (MPEG, AC-3) Data multiplexing (MPEG) Display technology (large wide screens) Transmission technology (modulation) Production Digital TV has Key Technologies that make it possible. Most production within the current TV stations already happens in the digital domain using standards such as Rec 601 digital video. It only becomes analog when it is transmitted over the air to the viewer. Display technology has not reached the level needed for HDTV to be fully implementable at present.

34 MPEG-2 Compresses source video, audio & data
Segments video into I, P & B frames Generates system control data Packetises elements into data stream Multiplexes program elements - services Multiplexes services - transport stream Organises transport stream data into 188 byte packets The functions of MPEG-2 are then as shown in this table.

35 Digital Terrestrial TV - Layers
. . . provide clean interface points Picture Layer Multiple Picture Formats and Frame Rates 1920 x 1080 1280 x 720 50,25, 24 Hz Video Compression Layer MPEG-2 compression syntax or Data Headers Motion Vectors Chroma and Luma DCT Coefficients Variable Length Codes Transport Layer MPEG-2 packets Video packet Audio packet Aux data Packet Headers Flexible delivery of data DTTB is about layers. Picture Compression Transport/Multiplexing Transmission Transmission Layer 7 MHz COFDM / 8-VSB VHF/UHF TV Channel

36 Digital Television Encode Layers
Picture Coding Audio Coding Data MPEG-2 or AC-3 MPEG-2 Control Video Sound Program 1 Multiplexer MPEG Transport Stream Mux Control Data (PMT) PES Bouquet Multiplexer Program 2 Program 3 Service Mux Other Data Control Data (PAT) Modulator & Transmitter Error Protection Control Data 188 byte packets MPEG Transport Data Stream What are the inputs to these layers? Delivery System

37 Digital Television Decode Layers
Data Mon Speakers Audio Decoder Data Decoder Picture MPEG or AC-3 MPEG-2 MPEG Transport Stream De-Multiplexer MPEG DeMux Transport Stream Demodulator & Receiver Error Control At the receiver we simply select the portions of the existing data stream we wish to decode and throw away the rest. Delivery System

38 Set top Box (STB) - Interfacing
Domestic and Professional interfaces still to be defined Most probably Transport Stream via IEEE 1394 (Firewire) Baseband Audio & RGB/YUV Video signals. STB can convert between line standards so you do not have to have a HD display. Display and transmitted information must be at same Frame/Field rate. (25/50)

39 DTTB - Content & Services
DTTB was designed to carry video, audio and program data for television DTTB can carry much more than just TV Electronic program guide, teletext Broadband multimedia data, news, weather Best of internet service Interactive services Software updates, games Services can be dynamically reconfigured DTTB can carry many other things than just television. Interactive services need a back channel such as the telephone line or a cable/wireless modem. The analog television we are used to uses a very dumb device to display the pictures. Digital TV uses a smart box which can be dynamically re-configured. You can choose the change the channel/data structure mid program, upload new operating software with different funtions. No longer will the transmission be totally constrained by the dumb receiver at the other end.

40 DVB Data Containers MPEG Transport Stream is used to provide DVB “data containers” which may contain a flexible mixture of: Video Audio Data services Streams with variable data rate requirements can be Statistically Multiplexed together. Allows Six 2 Mb/s programs to be placed in a 8 Mb/s channel

41 Examples of DVB Data Containers
Channel bandwidth can be used in different ways: SDTV 1 SDTV 2 SDTV 3 SDTV 4 SDTV 5 Multiple SDTV programs Single HDTV program HDTV 1 SDTV 1 HDTV 1 Simulcast HDTV & SDTV Focusing on the TV aspects. There are different ways the channel bandwidth can be used. As has been done in the UK, you can have a multiplex which has multiple standard definition programs within it. You can also have a multiplex which contains a single High Definition program which is the plan for many situations in Australia. There is also the facility to have simulcast, where you cut down a bit on the High Definition to allow the older receivers that are there to decode the Standard Definition variant. This may well be used in the future in the UK when the analog transmissions are turned off. The plan in most of Europe is for High definition to be introduced as a second phase as part of the analog turn off.

42 Video Program Capacity
For a payload of around 19 Mb/s 1 HDTV service - sport & high action 2 HDTV services - both film material 1 HDTV + 1 or 2 SDTV non action/sport 3 SDTV for high action & sport video 6 SDTV for film, news & soap operas However you do not get more for nothing. More services means less quality Digital TV will have a data capacity around 20 Mb/s. For Sport or high action we can have relatively few services. Films have high levels of temporal redundancy because both fields are scanned from the same frame. This allows the compression systems to perform higher levels of compression allowing spare data capacity and the ability to have more services. Generally News & Soapies have lower data requirements so more channels are possible. You do not ge more channels for nothing. More Services Means Less Quality.

43 Spare Data Capacity Spare data capacity is available even on a fully loaded channel. Opportunistic use of spare data capacity when available can provide other non real time data services. Example: 51 second BMW commercial The Commercial was shown using 1080 Lines Interlaced. 60 Mb of data was transferred during it. In the Final 3 seconds the BMW Logo was displayed allowing 3 Phone Books of data to be transmitted. An example from tests in America. Other non-real time services can use opportunistic use of the DTV data pipe to transmit data when the full bandwidth of the channel is not required for the main services. These services would be data specifically intended for broadcast application with no need for acknowledgement or a back channel. Teletext or Newspaper type information are good examples.


Download ppt "University of Canberra Advanced Communications Topics"

Similar presentations


Ads by Google