ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, GenevaInternational Telecommunication UnionPerceptual Encoder Perceptual Encoder OptimizationOptimizationDavid McNallyGenista Corp.2ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TCODECso Video codecs have reached a very high level of sophisticationo We have two mathematically “orthogonal” mainstream approaches:1. DCT with motion estimation & F/B prediction (MPEG family)2. Wavelet based schemes (JPEG2000)We observe ‘similar’ quality as function of target bit-rate & frame rate:3ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TCODECsDCT: Wavelet:CodingTransmission4ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TCODECso The currencies for achievable quality are:• Encoder usage: 1-pass vs. n-pass • Target Bitrate, frame size & frame rate• Network behavior (BER, PLR, Latency, Jitter)BLER vs.Perceived Quality5ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TCODECsEncoders have many(!) configuration parameters…! HOW TO CHOOSE OPTIMAL SETTINGS?6ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TCODECso CODECs are designed to handle ‘expected’ content well:• Sports; Movies; News; Nature; …o They fail for ‘unexpected’ content:Tape noise Tape drop-out7ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TCODECsEncoders are optimized for ‘mainstream’content! HOW TO HANDLE ‘OUTLIERS’?8ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TThe Bottom-Lineo The customer does not care what technology is usedo The customer has expectations which depend on his setting & experience:• IPTV / PPV / … ! Broadcast TV• Mobile / Wireless ! Internet streaming• Personalized ! High toleranceDoes the consumer feel he is getting good value for money?!9ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TThe Bottom-Lineo There is an ‘industry’ which has developed around codec performance assessments (e.g. VQEG):! (Subjective) Test methodologies! Quality algorithms! Statistical procedures! Vested interests10ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TSubjective testing010203040506070809010000:00:00:0000:04:00:1200:04:50:1200:05:40:1200:06:39:2400:07:29:2400:08:19:2400:27:19:1200:28:09:1200:28:59:1200:29:49:12TimeScores32 kbps128 kbps384 kbps17’ LCD monitorviewerdeskViewing room (*) viewing location 1 curtainwall backgroundviewing location 2 viewing location 3 Mean Opinion ScoreVariance(MOS)11ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TMethodologiesNo Reference: Measure subscriber perceived qualityReferenceProcess the contentProcessedMeasure QualityReferenceProcess the contentProcessedAlignmentMeasure QualityFull Reference: Measure relative degradation12ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TVideo Quality MetricsNo Reference:o Bluro Jerkinesso Blockinesso Colorfulnesso MOSFull Reference:o Bluro Jerkinesso Blockinesso Colorfulnesso Noiseo ANSI metricso PSNRo MOS13ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TNR-MOS: Predictive Power14ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TEncoder OptimizationReferenceParametersProcessedENCODERNRFRNR1: Technical optimization2: Service optimization3: Content optimization12315ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TThere is Really a Difference!NR BlurNR BlurNR BlockinessWindows MediaApple QuicktimeNR Blkss16ITU-T VICA Workshop22-23 July 2005, ITU Headquarter, Geneva datesITU-TConclusionso ‘Good’ Perceived quality is the key to successful media serviceso Encoders are mature and are reaching asymptotic performance limito Video content is highly variableo User expectations are highly variableo Provide cost-performance optimized services -Need to integrate:• Knowledge of content• Knowledge of user perception & expectation• Knowledge of encoder
View Full Document