(A.2) Theoretical Background of Data Compression 

When the general term of data is used here, main source of this data is image (gray or color), audio, video or both audio and video. Video is made of different frames (same as still image) passed in a sequence with 30 frames per second (fps) typically. Image, video, and audio data (signal) can be compressed due to these factors: 

I. There are generally three types of redundancies exist in image/video data as:

1. Spatial Redundancy or Spatial Correlation: 

Within the single image or single frame of video, there exists the significant correlation among the neighbor samples (pixels) and results spatial redundancy in data.

2. Spectral Redundancy or Spectral Correlation:

For the data acquired from multiple sources (e.g. multiple cameras), there exists significant correlation amongst samples from these sources. This correlation is referred as spectral correlation and results the spectral redundancy in data.

3. Temporal Correlation or Temporal Redundancy:

For temporal data (such as audio and video), there is significant correlation amongst samples in different segment of time. This is referred as temporal correlation and results the temporal redundancy in data.

The key idea in any compression is to reduce the number of highly correlated samples to lesser representative samples, in different group of correlated samples, which in turn results in reduced redundancy across the data and hence reduced size of data.

II. There is considerable information in the data that is irrelevant from a perceptual point      of view and discarding this information results the reduced amount of data.

III. Some data tends to have high-level features that are redundant across space and time; i.e. data is of a fractal nature.

 

(B) A Typical Block Diagram of E-Learning Classroom

 

Figure 1: A Typical Block Diagram of E-Learning Classroom

 

(C)  General Frameworks of Coder and Decoder 

 

Figure 2: Full Architecture of End to End Audio / Video Distribution and Storage Network

 

Description

I. The full end to end delivery system contains the major areas as:

1. Audio and Video Compression Module

2. Application Layer QoS control for Streaming Audio and Video

3. Continuous Media Distribution Service

4. Streaming Server Module

5. Media Synchronization Mechanism

6. Protocol for Streaming Media

7. Streaming Video over Wireless network Module

 

II. Typical constraints encounters during development of such system as:

1. Bandwidth of network

2. Delay in processing (both at network transportation level and processing delay at   sending and receiving ends)

3. Loss of data

4. Quality of data

III. Summary of functionality of different blocks (figure 2)

a. Video/Image Compression 

It has mainly two type of basic functionality to generate the compressed data, as :

• Scalable

• Non-scalable

b. Application layer QoS (Quality of Service) for streaming Video: This block is mainly responsible to cope the varying network conditions and ensures the presentation quality requested by user. It has mainly two functions, as:

• Congestion Control – to reduce the packet loss and reduce delay.

• Error Control – it improves the video presentation quality in the presence of packet loss.

c. Continuous Media Distribution Service: It is built on the top of network protocol used and designed to take support of network to reduce delay, packet loss and hence to achieve QoS and efficiency for audio, video transfer over the network. It mainly includes:

• Network Filtering

• Application level multicast

• Content Replication

d. Streaming Server: It processes multimedia data under timing constraints and support interactive control (e.g. pause, play, fast-forward, fast backward etc.). The proper working of streaming server requires that it must receive the media components in synchronous manner (at most very small time delay). It has mainly two parts:

• Communicator (transport protocol)

• Operating System

e. Media Synchronization Mechanism: It synchronizes the timing of audio and video streams while decoding to have faithful reconstruction at receiving side.

f. Protocols for Streaming Media: It can be of three types depending upon network:

• Network Layer Protocol (e.g. IP)

• Transport Protocol (e.g. UDP- User Datagram Protocol)

• Session Control Protocol (e.g. Real time Streaming Protocol – RTSP)

 

(D) General Coder Framework 

 

Figure 3: General Framework for Implementing Encoder / Decoder for Video

 

I. Two Classes of Algorithms to Implement Video Coding/Decoding Systems

• Waveform Based Video Coding  

   Allow the efficient coding of video without trying to analyze the video content.

• Content Dependent Video Coding

Allow the coding by analyzing the content of video i.e. algorithms identify the

                    regions and objects in a video sequence and code those.

II.  Model Adapted for Coding: There are some factors for model adapting the coding, as: 

1. The components of video coding algorithms are determined to a large extent by the source model adapted for modeling of video sequences.

2. The video coder seeks to describe the contents of video sequences by means of its source model.

3. The source model may make assumption about the spatial and temporal correlation between pixels of a sequence. It might also consider the shape and motion of objects or illumination effects.

III. Description of Blocks Used (figure 3)

a. Source Model: Video Modeling Technique and based on parameters of source model, coder seeks to      describe the content of video sequence.

b. Analysis:  Digitized raw video described using parameters of source model.

• Considering pixels are statistically independent:

Parameters of source model would be ‘luminance’ and ‘chrominance’ amplitude of each pixel.

• Considering correlated pixels (scene as several objects):

Parameters of source model would be shape, texture and motion of individual objects.

c. Quantization: Parameters of source model are quantized into finite set of symbols depending upon constraint on bit-rate and distortion. 

d. Binary Encoding: Quantized parameters are mapped to binary code words using lossless coding technique.                            

  

(E) General Model of Layered Video Coder

 Figure 4: Block Diagram Layered Video Codec

 

Description of Different Layers (figure 4):

Layer 1: Transmits only the color information (parameters) of   statistically independent pixels (corresponds to ‘I’ frame of hybrid coder).

Layer 2: Allows the additional transmission of motion parameters for the block of fixed size and position (corresponds to ‘P’ and ‘B’ frame coder of hybrid coder).

Layer 3: Allows to transmission of shape information in order to describe arbitrarily shaped objects.

Layer 4: The knowledge of scene contents is extracted from video sequence and transmitted in order to apply scene dependent object model.

Layer 5: Transmits the high level symbols for describing complex behaviors of objects.

Layer Selector: This block selects the layer depending upon the object properties, video quality needed, bit rate, coding error etc.