International Telecommunication Union Perceptual Encoder Optimization David McNally Genista Corp ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva CODECs ITU T o Video codecs have reached a very high level of sophistication o 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 dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 2 CODECs DCT Wavelet Transmission Coding ITU T dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 3 CODECs o ITU T 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 Quality dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 4 CODECs ITU T Encoders have many configuration parameters HOW TO CHOOSE OPTIMAL SETTINGS dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 5 CODECs o ITU T CODECs are designed to handle expected content well o Sports Movies News Nature They fail for unexpected content Tape noise dates Tape drop out ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 6 CODECs ITU T Encoders are optimized for mainstream content HOW TO HANDLE OUTLIERS dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 7 The Bottom Line ITU T o The customer does not care what technology is used o The customer has expectations which depend on his setting experience IPTV PPV Broadcast TV Mobile Wireless Internet streaming Personalized High tolerance Does the consumer feel he is getting good value for money dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 8 The Bottom Line ITU T o There is an industry which has developed around codec performance assessments e g VQEG dates Subjective Test methodologies Quality algorithms Statistical procedures Vested interests ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 9 Subjective testing wall background 17 LCD monitor desk ITU T viewer viewing location 2 viewing location 1 viewing location 3 curtain Viewing room 100 90 80 70 Mean Opinion Score Scores 60 50 32 kbps 128 kbps 384 kbps 40 30 20 10 00 29 49 12 00 28 59 12 00 28 09 12 00 27 19 12 00 08 19 24 00 07 29 24 00 06 39 24 00 05 40 12 00 04 50 12 00 04 00 12 00 00 00 00 0 Time Variance MOS dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 10 Methodologies No Reference Measure subscriber perceived quality ITU T Reference Process the content Processed Measure Quality Full Reference Measure relative degradation Reference Process the content Processed Alignment Measure Quality dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 11 Video Quality Metrics No Reference ITU T o o o o Blur Jerkiness Blockiness Colorfulness Full Reference o Blur o Jerkiness o Blockiness o Colorfulness o Noise o ANSI metrics o PSNR o MOS dates o MOS ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 12 NR MOS Predictive Power ITU T dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 13 Encoder Optimization FR ITU T 1 Parameters ENCODER Reference NR 2 Processed NR 3 1 Technical optimization 2 Service optimization 3 Content optimization dates ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 14 dates Windows Media ITU T NR Blur Apple Quicktime There is Really a Difference NR Blur NR Blockiness NR Blkss ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 15 Conclusions ITU T o o o o o Good Perceived quality is the key to successful media services Encoders are mature and are reaching asymptotic performance limit Video content is highly variable User expectations are highly variable Provide cost performance optimized services Need to integrate dates Knowledge of content Knowledge of user perception expectation Knowledge of encoder implementation ITU T VICA Workshop 22 23 July 2005 ITU Headquarter Geneva 16
View Full Document
Unlocking...