12.07.2015 Views

Scalable H.264 Video Coding Using MCTF - SIPL

Scalable H.264 Video Coding Using MCTF - SIPL

Scalable H.264 Video Coding Using MCTF - SIPL

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Students:Supervisors:Alexander ShamaevGregory VaksmanHilla MadarYair Moshe1<strong>SIPL</strong> Annual Projects presentation, 2008


ScalabilityGoals and usage<strong>Scalable</strong> coding offers multiple levels of decoding:Reciever1<strong>Scalable</strong>CoderBase layerEnhancement LayerReciever22Base-layer provides coarse reconstructionEnhancement-layer improves the quality


ScalabilityVarious types of scalability• Spatial scalability• Temporal scalability• Quality (PSNR) scalability3


ScalabilityAn example of classical scalability schemeA two level spatially/temporally scalable codec encoder:Raw<strong>Video</strong>DCTQVLCBase LayerCompressedBit-stream= = Down-sampling= Up-samplingIDCTDCTIQQVLCEnhanced LayerCompressedBit-streamVLC – variable-lengthcodePrimary Disadvantage:High bit rate4


Scalability in <strong>H.264</strong>Concepts of Motion Compensated Temporal Filtering (<strong>MCTF</strong>)• Based on the lifting scheme– Ensures perfect reconstruction (withoutquantization)– Even if non-linear operations are used• Hierarchical prediction structure• Preventing drift error5


Scalability in <strong>H.264</strong>Lifting Scheme2S2k+ 1H 2kS '2 k + 1S kS ' kPUUP2S 2kAnalysis Filter BankL kS '2 k2Synthesis Filter BankP = PredictionU = Update= Down-sampling= Up-sampling<strong>MCTF</strong> is innovative because for Predictionand Update it uses the standard motionestimation of an <strong>H.264</strong> encoder.6


Scalability in <strong>H.264</strong>Analysis Filter BankS2k2k1S +kL kH7


Scalability in <strong>H.264</strong>LkSynthesis Filter BankHk'S 2k'S2k+ 18


Scalability in <strong>H.264</strong>Filter BanksS 4kS4k+ 1S4k+ 2S 4k+ 3L 2H 1H 2kL 1L2kL2k+ 1H2k+ 1H1LkHkL 0H 09


Scalability in <strong>H.264</strong>Iterative temporal decomposition by <strong>MCTF</strong>10The red-framed pictures need to be transmitted together with thecorresponding motion information


Scalability in <strong>H.264</strong>Comparison resultsContainer CIFForeman CIF11


Summary‣ For many application scenarios the requirement forscalable video coding techniques is evident‣ <strong>MCTF</strong> extension of <strong>H.264</strong>/AVC provide someadvantages in coding efficiency compared to previoussolutions‣ The prediction methods used for spatial scalability,are subject to future refinements12– For slow motion sequences, scalable encoding isfairly better than ‘regular’ encoding– For fast motion sequences, ‘regular’ encoding isslightly better than scalable encoding, but the bit ratepenalty is not big

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!