A hardware-Friendly Wavelet Entropy Codec for Scalable video

1
A hardware-Friendly Wavelet Entropy Codec for
Scalable video

• Hendrik EeckhautELIS-PARISGhent
  UniversityBelgium

2
Scalable video
Decode according torequired QoS oravailable
hardware resources.
Encode once
Quality deployed hardware resources
3
Overview

• Scalable video codec
• Scalable Wavelet Entropy Encoding
• New algorithm
• Memory footprint
• Compression results
• Conclusions

4
Overview video codec
Exploit temporalredundancy
Exploitspatial redundancy
Exploit statistical properties
5
Motion estimation
? vectors error frame
6
Motion compensation
Reference framecurrent GOP
Reference framenext GOP
time
Scalability in time
7
Wavelet transform
Scalability in resolution
8
Wavelet Entropy Coder

• Actual compression of frames
• Tight coupling of pixel characterization and
  Arithmetic coding (modified arithmetic encoder of
  CABAC)

• Scalability
• Quality
• Resolution
• Good compression
• Economical with memory
• High degree of parallelism
• Sweet and simple

Wavelet Entropy Codec
9
Scalability in quality
Scalability in quality
10
Parallelism
inParallel
In parallel
Easy to omit subbands ? resolution scalability
11
Algorithm
12
Algorithm
Significance bitmap
Bitplane
Sign bitmap
13
Context/Arithmetic models

• 64 different models each model needs only a 9
  bit state
• 1 data model
• 27 sign models
• 27 significance models
• 8 special highest bitlayer significance models
• 1 refinement model

• Models are initialised (warmed up)with values
  obtained from training a number of representative
  sequences

14
Memory requirements
Altera Stratix S25
Name Number Size (byte parity)
MRAM 2 64k
M4K 138 512
M512 224 64
Registers 25660 1/8
214kiB

• Arithmetic Decoder
• State Lookup table 256B
• Range Lookup table 2048B
• Model state table 64B
• Buffer
• Model Selector
• Significance and sign bitmap 2 x 3168B
• A few buffers
• (1 MRAM block is large enough to contain
  multiplesignificance and sign bitmaps)

15
Results
QTL QuadTree Limited (comparable compression as
EBCOT of jpeg2000)
16
Conclusion

• Algorithm is simple
• No recursive behaviour
• Small memory footprint 2 bitmaps of
  size(wavelet subband)
• Memory access is very regular
• Arithmetic coder must process a lot symbols, but
  is simple and fast
• Compression is better than QTL-algorithm for all
  bitrates

17
(No Transcript)
18
Arithmetic Coding
19
Scalable video (animated)
Decode according torequired QoS oravailable
hardware resources.
Encode once
Quality deployed hardware resources
20
Overview video codec
MotionEstimation
EntropyEncoding
WaveletTransform
Pack
Motion VectorEncoding
originalframes
Pullbitstream
Unpack
MotionComp.
EntropyDecoding
InverseWavelet T.
Motion VectorDecoding
decodedframes