Layered Coding

Basic Overview

Outline

Pyramidal Coding

Scalability in the Standard Codecs

Layered Coding with Wavelets

Conclusion

Laplacian w/ Gaussian Pyramids

Laplacian w/ DCT Coefficient Pyramids

DCT Pyramid Example

Figure 7.5

Reconstruction Data for Fig. 7.5

Scalability in the Standard Codecs

Data Partitioning

Signal-to-Noise Ratio (SNR)

Spatial

Temporal

Data Partitioning

SNR Scalable Coder

SNR Scalability – no drift

Spatial Scalability

Temporal Scalability

Scalability Applications

Data Partitioning– Video of ATM networks

SNR– Transmission of video at different qualities

Video on demand, simultaneous TV and enhanced TV

– Video over high packet loss networks (Internet)

Spatial Scalability– Internetworking between two different standards– Simulcasting of drift-free video (TV & HDTV)

Scalability Applications

Spatial Scalability (continued)– Reception of low spatial resolution pictures over

mobile networks

Temporal Scalability– Migration to HDTV from current interlaced– Internetworking between mobile and fixed networks

Layered Coding with Wavelets

Growing interest due to new efficient coding techniques (Embedded Zero-tree Wavelets, or EZW)

Accepted for coding of still images– MPEG-4 and JPEG 2000

Predicted to continues in growth

Discrete Wavelet Transform

Split signal spectrum into several frequency bands analysis filters

Figs. 7.27, 7.28, 7.29

Distortions

Filter Bank Solutions

Higher Order Systems

Multiple wavelet transform coding by means of repeated two-band splits

Further Considerations

• Wavelet Example – Fig. 7.32

• Zero-Tree Coding – Fig. 7.34

• Quad-Tree Representation- Bands look like scaled versions of each other- Lower bands have 0.25 dimension of higher

Conclusion

Layered coding facilitates– Error protection– Efficient use of resources– Multiple services– Customer satisfaction