Nippon Telegraph and Telephone Corporation (NTT), Japan Taichi Kawano.

Post on 17-Jan-2016

221 views 0 download

Tags:

Transcript of Nippon Telegraph and Telephone Corporation (NTT), Japan Taichi Kawano.

Nippon Telegraph and Telephone Corporation (NTT), Japan

Taichi Kawano

Conducted subjective quality assessment for ACR, DCR, and DSCQS to compare their stabilities in 3D video

Experimental results showed that ACR is suitable in terms of its stability and assessing time

ACR

0.22

0.26

0.30

0.34

0.38

0.42

0.46

12 16 20 24 28 32 36 40 44

MC

I

M

3D2D

MCI of ACR for 3D video was larger than that for 2D video.

DCR

0.220.260.300.340.380.420.46

12 16 20 24 28 32 36 40 44

MC

I

M

3D2D

456789

10

12 16 20 24 28 32 36 40 44

MC

I

M

3D2D

DSCQS

No significant difference in MCI of DCR and DSCQS between 2D and 3D videos.

In DCR and DSCQS, it might be easy for participants to evaluate 3D video quality because there is a reference video

In ACR, it might be difficult for participants to evaluate 3D video quality because they do not have a criterion for evaluating 3D video quality due to lack of viewing experience

0.06

0.09

0.12

0.15

12 16 20 24 28 32 36 40 44

Nor

mal

ized

MC

I

M

ACRDCRDSCQS

minmax MOSMOS

MCI MCI Normalized

0.06

0.09

0.12

0.15

12 16 20 24 28 32 36 40 44

Nor

mal

ized

MC

I

M

ACRDCRDSCQS

Normalized MCI of ACR with 24 participants for 2D video

Normalized MCI of ACR with 24 participants for 2D video

To satisfy the criterion (Normalized MCI < 0.09)

28 people must participate in ACR

23 people must participate in DCR

36 people must participate in DSCQS

DCR is suitable in term of MHowever, the assessing time of DCR is twice as long as that of ACR

Assessing time of ACR per participant is T

If “Normalized MCI < 0.09” is neededACR requires 28T (T * 28 people)DCR requires 46T (2T * 23 people)DSCQS requires 144T (4T * 36 people)

ACR is suitable in term of its stability and assessing time.

Propose using ACR with more than 28 participants for 3D video

Scores of 42 participants

P1 : Score=4P2 : Score=5

P42 : Score=5

Scores of M participants

P1 : Score=4P2 : Score=5

P42 : Score=5

Randomly select

CalculateMCI

MCI

Selection and calculation were repeated 15 times

MCI in graph plots average of 15 MCIs