evaluation of mpeg4-svc f or qoe protection i n the context of transmission errors

17
Evaluation of MPEG4-SVC for QoE protection in the context of transmission errors Yohann Pitrey Marcus Barkowsky Patrick Le Callet Romuald Pepion IRCCyN lab, IVC group NANTES, FRANCE Ecole Polytechnique de l’Universite de Nantes SPIE Optical Engineering + Applications Applications of Digital Image Processing XXXIII

Upload: shaquille-charles

Post on 01-Jan-2016

11 views

Category:

Documents


1 download

DESCRIPTION

Yohann Pitrey Marcus Barkowsky Patrick Le Callet Romuald Pepion. Evaluation of MPEG4-SVC f or QoE protection i n the context of transmission errors. SPIE Optical Engineering + Applications Applications of Digital Image Processing XXXIII. IRCCyN lab, IVC group NANTES, FRANCE - PowerPoint PPT Presentation

TRANSCRIPT

Evaluation of MPEG4-SVC for QoE protection in the context of transmission errors

Yohann Pitrey

Marcus Barkowsky

Patrick Le Callet

Romuald PepionIRCCyN lab, IVC group

NANTES, FRANCE

Ecole Polytechnique de l’Universite de Nantes

SPIE Optical Engineering + ApplicationsApplications of Digital Image Processing XXXIII

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 2 -

Scalable Video Coding for error concealment

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 3 -

Outline

Workflow

SVC-based error concealment

Subjective experiment setup

Experimental results

Conclusions

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 4 -

Workflow overview

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 5 -

SVC-based error concealment

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 6 -

Subjective experiment setup (1/2)

Viewing conditions 40-inch reference full-HD LCD screen 1 session of 40 min. 145 sequences of 10 sec. Viewing distance = 4 x screen height (2m)

Assessment methodology Absolute Category Rating (ACR)

with 5 levels (bad …. excellent)

Observers 29 non-expert viewers Visual acuity / color blindness tested

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 7 -

Subjective experiment setup (2/2)

Tested video sequences

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 8 -

Experimental results (1/3)

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 9 -

Experimental results (2/3) *

refe

ren

ce

Non

dam

aged

SV

C

Dam

age

d A

VC

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

120 kbps15 fps

120kbps30 fps

200 kbps15 fps

200 kbps30 fps

- _ _ _ _

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 10 -

refe

ren

ce

Non

dam

aged

SV

C

Dam

age

d A

VC

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

120 kbps15 fps

120kbps30 fps

200 kbps15 fps

200 kbps30 fps

Experimental results (2/3) *

_ - _ _ _

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 11 -

refe

ren

ce

Non

dam

aged

SV

C

Dam

age

d A

VC

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

120 kbps15 fps

120kbps30 fps

200 kbps15 fps

200 kbps30 fps

Experimental results (2/3) *

_ _ - _ _

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 12 -

refe

ren

ce

Non

dam

aged

SV

C

Dam

age

d A

VC

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

120 kbps15 fps

120kbps30 fps

200 kbps15 fps

200 kbps30 fps

Experimental results (2/3) *

_ _ _ - _

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 13 -

refe

ren

ce

Non

dam

aged

SV

C

Dam

age

d A

VC

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

upsc

ale

patc

h

switc

h

120 kbps15 fps

120kbps30 fps

200 kbps15 fps

200 kbps30 fps

Experimental results (2/3) *

_ _ _ _ -

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 14 -

Experimental results (3/3)

Statistical verification (student t-test)

a

b

a

b

a

b

a statistically better than b

a and b not statistically significantly

different

b statistically better than a

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 15 -

Experimental results (4/4)

Yohann Pitrey - MPEG-4 SVC for error concealment – 08.03.2010- 16 -

Conclusions

SVC + concealment gets better quality than AVC + concealment

Spatial discontinuities less visible than temporal discontinuities Layers with equal FPS => pixel-level concealment Layers with different FPS => frame-level concealment

Thank you for your attention

Your questions are welcome!