No-Reference Metrics No-Reference Metrics For Video Streaming For Video Streaming Applications Applications International Packet Video Workshop (PV 2004) International Packet Video Workshop (PV 2004) Presented by : Bhavana Presented by : Bhavana CPSC 538 CPSC 538 February 21, 2004 February 21, 2004
26
Embed
No-Reference Metrics For Video Streaming Applications International Packet Video Workshop (PV 2004) Presented by : Bhavana CPSC 538 February 21, 2004.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
No-Reference Metrics For No-Reference Metrics For Video Streaming ApplicationsVideo Streaming Applications
International Packet Video Workshop (PV 2004) International Packet Video Workshop (PV 2004)
What is a No-Reference metric ?What is a No-Reference metric ?
Estimating end-user’s QoE of a Estimating end-user’s QoE of a multimedia stream without using an multimedia stream without using an original stream as a reference.original stream as a reference.
In other words :In other words :
“ “ Quantify quality via blind distortion Quantify quality via blind distortion measurement”measurement”
PurposePurpose
To evaluate two types of distortions in To evaluate two types of distortions in streaming of compressed video over streaming of compressed video over packet-switched networkspacket-switched networks
- Compression related : block-edge - Compression related : block-edge impairment impairment
- Transmission related : packet-loss - Transmission related : packet-loss impairment impairment
Where Can It Be Used ?Where Can It Be Used ?
For real time monitoring .For real time monitoring .
Reference unavailable or expensive to sendReference unavailable or expensive to send
Feedback to Streaming Server .Feedback to Streaming Server .
Evaluation of Compression AlgorithmsEvaluation of Compression Algorithms
What are Block-Based Codecs ?What are Block-Based Codecs ?
Process several pixels of video together Process several pixels of video together in blocksin blocks At high compression rates, strong At high compression rates, strong discontinuities called block edges come discontinuities called block edges come up.up.What’s blockiness ? What’s blockiness ?
“ “Distortion of image characterized by Distortion of image characterized by appearance of underlying block encoding appearance of underlying block encoding structurestructure
Block – based DistortionBlock – based Distortion
Idea Idea : A block-edge gradient can be : A block-edge gradient can be masked by a region of high spatial activity masked by a region of high spatial activity around it .around it .
Measure two things :Measure two things :
- spatial activity around block edges : - spatial activity around block edges : σσ
Bit Rate => compression levelBit Rate => compression level
NR Packet Loss MetricNR Packet Loss Metric
Error Concealment : Replace Error Concealment : Replace damaged/lost macroblock with damaged/lost macroblock with corresponding macroblock from previous corresponding macroblock from previous frame.frame.
Idea Idea : Use length of artifact to estimate : Use length of artifact to estimate amount of distortion caused by packet lossamount of distortion caused by packet loss
Calculation of NR Packet Loss Calculation of NR Packet Loss MetricMetric
For a m x n frameFor a m x n frame
For each 16 x 16 macroblockFor each 16 x 16 macroblock
Calculate :Calculate :
ÊÊjj = strength vector across macroblock = strength vector across macroblock
edgeedge
ÊÊ΄́jj = strength vector within macroblock = strength vector within macroblock
near the edgenear the edge
Macroblock 1
Macroblock 2
Figure : Calculating Strength vector across and within a macroblock
Convert strength vectors into binary Convert strength vectors into binary vectorsvectors
EEjj(k) = 1 if (k) = 1 if ÊÊj j > > ττ
= 0 otherwise= 0 otherwise
EE΄́jj(k) = 1 if (k) = 1 if ÊÊ΄́j j > > ττ
= 0 otherwise= 0 otherwise
where where ττ = 15 = 15
If the sum of differences between the two binary If the sum of differences between the two binary edge vectors is substantial , then there is edge vectors is substantial , then there is distortiondistortion
Packet loss metric for jPacket loss metric for j thth macroblock macroblock
where where ζζ = 10% of frame width (n) = 10% of frame width (n)
Packet loss metric for whole frame Packet loss metric for whole frame
F = ∑ HF = ∑ Hjj22
Simulation Setup for NR Packet Simulation Setup for NR Packet Loss MetricLoss Metric
Bit Rate = 1.5 MbpsBit Rate = 1.5 Mbps
Frame Rate = 30 fpsFrame Rate = 30 fps
Frame Size = 352 x 240Frame Size = 352 x 240
Used NTT DoCoMo packet loss Used NTT DoCoMo packet loss generating software .generating software .
Limitations Of NR-metricsLimitations Of NR-metrics
Blockiness metric might fail in the Blockiness metric might fail in the presence of strong de-blocking filters presence of strong de-blocking filters which might otherwise introduce blurwhich might otherwise introduce blur
Metric predictions lose meaning in Metric predictions lose meaning in presence of other distortions like blur, presence of other distortions like blur, noise etc.noise etc.
Future DirectionsFuture Directions
VQEG standardization effortsVQEG standardization efforts HVS based approachesHVS based approaches Statistical models for natural scenesStatistical models for natural scenes NR QA schemes for NR QA schemes for
- Non-block based compression schemes - Non-block based compression schemes such Wavelet-based such Wavelet-based
-Targeting full range of artifacts -Targeting full range of artifacts
ReferencesReferences
No Reference Image and Video Quality No Reference Image and Video Quality AssessmentAssessment http://http://live.ece.utexas.edu/research/quality/nrqa.htmlive.ece.utexas.edu/research/quality/nrqa.htm Objective video Quality Assessment Objective video Quality Assessment http://www.cns.nyu.edu/~zwang/files/papers/QAhttp://www.cns.nyu.edu/~zwang/files/papers/QA_hvd_bookchapter.pdf_hvd_bookchapter.pdfPerceptual Video Quality and Blockiness Metrics Perceptual Video Quality and Blockiness Metrics for Multimedia Streaming Applicationsfor Multimedia Streaming Applicationswww.stefan.winkler.net/Publications/www.stefan.winkler.net/Publications/wpmc2001.pdf wpmc2001.pdf