A new methodology to estimate the impact of H.264 artefacts on subjective video quality

57 %
43 %
Information about A new methodology to estimate the impact of H.264 artefacts on...
Technology

Published on November 19, 2008

Author: stephanepechard

Source: slideshare.net

Description

Presentation of my scientific paper to the Third International Workshop on Video Processing and Quality Metrics (VPQM2007).

A new methodology to estimate the impact of H.264 artefacts on subjective video quality Stéphane Péchard, Patrick Le Callet, Mathieu Carnec, Dominique Barba Université de Nantes – IRCCyN laboratory – IVC team Polytech’Nantes, rue Christian Pauc, 44306 Nantes, France Third International Workshop on Video Processing and Quality Metrics for Consumer Electronics Scottsdale, 2007-01-26

Introduction • Codec => Coding artefacts • Quality loss due to artefacts => Useful for quality metrics or better coding … • Possible practical approach – Artefact classification – Annoyance or quality loss contribution per artefact type 2

Farias and al. methodology Farias VPQM 05 • Artefacts type set (blockiness, blur, ringing, …) • Generation of synthetic artefacts – Strength parameter – Applied with the same strength on a whole part of the sequence • Subjective assessment => Annoyance curve per artefact type regarding the strength => Content dependency Alternative approach : VPQM07 ⇒ H.264 coding, Subjective assessment : quality scale, blur scale, blockiness scale … 3 ⇒ No direct control of artefacts strength

Proposed approach • H.264 artefacts due to quantization/decision • effects are different regarding the local content (edge, texture, …) • different perceived annoyance depending on the local spatio-temporal activity of the content • H264 distortions only in selected coherent spatio- temporal regions => define content categories • Subjective quality assessment ⇒ Quality loss curve per local content category (e.g. effects of H264 on each category) ⇒ Strength ? 4

Outline • Spatio temporal segmentation • distorted sequences generation • subjective quality assessment of sequences • Quality assessment : Combining categories • Towards quality loss function per content category 5

The approach source temporal unlabeled borders classification segmentation holes filling processing categories masks sequence C-distorted …… H.264 coding sequences Ci generation partly-distorted sequences usable for subjective tests 6

Spatio temporal classification 2 steps - temporal segmentation : reliability regarding the motion => temporal tubes - tube classification : Regarding spatial content 7

Segmentation of sequences source temporal unlabeled borders classification segmentation holes filling processing Class-distorted …… H.264 coding sequences Ci generation partly-distorted sequences usable for subjective tests 8

Segmentation of sequences • per group of five successive frame, the center frame is divided into blocks • motion estimation of each block using the two previous frames and the two next frames (motion estimation performed on a multi-resolution representation) i-2 i-1 i i+1 i+2 9

Segmentation of sequences • temporal tracking of each block of frame i defines a spatio-temporal “tube” over the five frames • a tube is oriented along the local motion i-2 i-1 i i+1 i+2 10

Classification source temporal unlabeled borders classification segmentation holes filling processing C-distorted …… H.264 coding sequences Ci generation partly-distorted sequences usable for subjective tests 11

Definition of content categories HVS has different perception of impairments depending on the local spatio-temporal content. • low luminance smooth areas; • high luminance smooth areas; • fine textured areas; • edges; • strong textured areas 12

Classification • 4 spatial gradients means per tube (directions : 0, 90, 45 and 135°) • plot in spatial space P (0 and 90°) => C1, C2, C3 and C4 • 2nd step : space P’ (45 and 135°) used to discriminate C5 in P • frontier determined to obtain relevant classification 13

Classification • global tracking of moving objects over the whole sequence • tubes are classified then merged by categories smooth areas with low luminance smooth areas with high luminance fine textured areas edges strong textured areas 14

Unlabeled holes filling and tube intersections source temporal unlabeled borders classification segmentation holes filling processing C-distorted …… H.264 coding sequences Ci generation partly-distorted sequences usable for subjective tests 15

Unlabeled holes filling and tube intersection • every pixel of the source has one and only one label • unlabeled holes : – gradient value => class – closest tube • Insection pixels : same 16

Borders processing source temporal unlabeled borders classification segmentation holes filling processing C-distorted …… H.264 coding sequences Ci generation partly-distorted sequences usable for subjective tests 17

Borders processing • borders between original and distorted large regions are treated so as to smooth the transitions before after borders borders processing processing 18

H.264 coding and class-distorted sequences generation source temporal unlabeled borders classification segmentation holes filling processing category masks sequence C-distorted …… H.264 coding sequences Ci generation partly-distorted sequences usable for subjective tests 19

Partly-distorted sequences generation original sequence C1 C2 H.264 sequences at different bitrates C3 C4 C5 categories sequence 20 5 sequences per bitrate

Original sequence (first frame) 21

One caregory distorted sequence (first frame) 22

Subjective quality assessment • SAMVIQ protocol with at least 15 validated observers and normalized conditions • 1920x1080 HDTV Philips LCD display • Doremi V1-UHD 1080i HDTV player 23

Subjective quality assessment • 11 sequences in a SAMVIQ session: – 5 Ci-only distorted at a certain bitrate B – entirely distorted sequence at B – entirely distorted sequence at low bitrate – entirely distorted sequence at intermediate bitrate – explicit and hidden references 24

Sequences uncompressed HDTV sequences from SVT Above marathon Captain Dance in the woods Duck fly C5 50 % C2 78 % C3 54 % C5 60 % Fountain man Group disorder Rendezvous Ulriksdals C2 71 % C2+C3+C1 95 % C5 56 % C2+C3 80 % 25

example on sequence Ulriksdals coded at 1 Mbps 90 80 70 60 50 40 30 20 10 0 1 2 3 4 5 Classes MOS(Sj,Bk) MOSref 26

DMOS and ∆MOS MOSref • MOS(Ci, Sj ,Bk) for each sequence ∆MOS(C4) Sj, each category Ci at each bitrate Bk MOS(C4) ∆MOS(C5) • DMOS(Sj ,Bk) = MOSref – MOS(Sj ,Bk) is the quality difference between MOS(C5) the reference and the entirely ∆MOS(C3) distorted sequence DMOS(Sj,Bk) MOS(C3) • ∆MOS(Ci, Sj ,Bk) = MOSref - MOS(Ci, ∆MOS(C1) MOS(C1) Sj ,Bk) is the quality loss induced by distortions in category Ci MOS(C2) ∆MOS(C2) MOS(Sj,Bk) 27

Possible relation between global DMOS and category ∆MOS? Combination CC ∆MOS(C2)+ ∆MOS(C4) + ∆MOS(C5) 0.9485 ∆MOS(C2) + ∆MOS(C5) 0.9440 • relations use ∆MOS(C2) + ∆MOS(C3) + ∆MOS(C4) 0.9094 ∆MOS(C1) + ∆MOS(C2) + ∆MOS(C3) sums of ∆MOS 0.9058 + ∆MOS(C4) + ∆MOS(C5) … … ∆MOS(C2) 0.7664 ∆MOS(C3) 0.7094 ∆MOS(C5) 0.6400 ∆MOS(C4) 0.5472 ∆MOS(C1) 0.5349 28

Non linear functions • DMOSp = maxi(∆MOSi) – CC = 0.9467 • DMOSp = maxi(∆MOSi) + maxj(∆MOSj) with j≠i – CC = 0.9530 • Correlation exists between global DMOS and category ∆MOS => DMOS could be predicted from quality per category 29

Towards a quality loss model • How to control the distortion level of a given class ? – Farias approach :strength of synthetic artefact • Factors implied in the quality loss of category Ci: – distortions themselves – motion – proportion of the category – spatial localisation (not considered here) 30

Distortion strength for category C1 • distortion strength = f(M,P,E) With all along the sequence : – M the mean motion of the category; – P the mean proportion of the category; – E the MSE on the category; • M decreases the distortion strength while P and E increase DS proposed model for f DS = (1 — M/Mt)×P×E 31

Quality loss function for category C1 • Psychometic function as a prediction of ∆MOS1 φ(DS) = (a×DSb)/(c+DSb) • correlation between φ(DS) and ∆MOS1 : 0.9514 • RMSE = 5.25 • good predictor of the loss of quality induced by category C1 32

Quality loss function for class C1 33 => Possible prediction of ∆MOS1

Conclusion • design of a new methodology to estimate the impact of H.264 artefacts on subjective video quality • One distortion type but – Effect related to local content – possibility to relate the global loss to loss per category – quality loss function for category C1 • Other categories and objective models 34

Results: segmentation statistics Above marathon Captain Dance in the woods Duck fly Séquence C1 (%) 3.75 13.14 3.80 0.13 C2 (%) 17.45 78.26 22.57 8.97 C3 (%) 27.79 6.81 53.85 19.50 C4 (%) 0.94 1.43 3.02 10.70 C5 (%) 50.06 0.36 16.75 60.70 35

Results: segmentation statistics Fountain man Group disorder Rendezvous Ulriksdals Séquence C1 (%) 10.52 25.28 8.78 13.54 C2 (%) 70.71 38.58 12.38 41.31 C3 (%) 13.37 29.80 19.87 40.48 C4 (%) 1.45 1.79 2.05 1.36 C5 (%) 3.93 4.54 56.92 3.30 36

Add a comment

Related presentations

Related pages

A new methodology to estimate the impact of H.264 ...

A NEW METHODOLOGY TO ESTIMATE THE IMPACT OF H.264 ARTEFACTS ON SUBJECTIVE VIDEO QUALITY Stéphane Péchard, Patrick Le Callet, Mathieu Carnec, Dominique Barba
Read more

[hal-00275334, v1] A new methodology to estimate the ...

A NEW METHODOLOGY TO ESTIMATE THE IMPACT OF H.264 ARTEFACTS ON SUBJECTIVE VIDEO QUALITY Stéphane Péchard, Patrick Le Callet, Mathieu Carnec, Dominique Barba
Read more

A new methodology to estimate the impact of H. 264 ...

A NEW METHODOLOGY TO ESTIMATE THE IMPACT OF H.264 ARTEFACTS ON SUBJECTIVE VIDEO QUALITY Stéphane Péchard, Patrick Le Callet, Mathieu Carnec, Dominique ...
Read more

A new methodology to estimate the impact of H.264 ...

... A new methodology to estimate the impact of H.264 ... H.264 artefacts on subjective video quality. ... H.264 artefacts on subjective quality ...
Read more

A new methodology to estimate the impact of H.264 ...

A new methodology to estimate the impact of H.264 artefacts on subjective video quality (2007)
Read more

stephane Péchard | Papers - Academia.edu

A new methodology to estimate the impact of H. 264 artefacts on subjective video quality more. ... A new methodology to estimate the impact of H. 264 ...
Read more

stephane Péchard - Academia.edu

A new methodology to estimate the impact of H. 264 artefacts on subjective video quality more. by stephane Péchard. Publication Date: 2007.
Read more

Video Quality | LinkedIn

Video Quality Engineer and Architect at eBrisk Video Past Embedded Video Codec Engineer (Telnet contractor) at Thomson Video Networks, Video Codec & ...
Read more