Three-dimensional (3D) video service is expected to be introduced as a next-generation television service. Stereoscopic video is composed of two 2D video signals for the left and right views, and these 2D video signals are encoded. Video quality between the left and right views is not always consistent because, for example, each view is encoded at a different bit rate. As a result, the video quality difference between the left and right views degrades the quality of stereoscopic video. However, these characteristics have not been thoroughly studied or modeled. Therefore, it is necessary to better understand how the video quality difference affects stereoscopic video quality and to model the video quality characteristics. To do that, we conducted subjective quality assessments to derive subjective video quality characteristics. The characteristics showed that 3D video quality was affected by the difference in video quality between the left and right views, and that when the difference was small, 3D video quality correlated with the highest 2D video quality of the two views. We modeled these characteristics as a subjective quality metric using a training data set. Finally, we verified the performance of our proposed model by applying it to unknown data sets.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Kazuhisa YAMAGISHI, Taichi KAWANO, Takanori HAYASHI, Jiro KATTO, "Subjective Quality Metric for 3D Video Services" in IEICE TRANSACTIONS on Communications,
vol. E96-B, no. 2, pp. 410-418, February 2013, doi: 10.1587/transcom.E96.B.410.
Abstract: Three-dimensional (3D) video service is expected to be introduced as a next-generation television service. Stereoscopic video is composed of two 2D video signals for the left and right views, and these 2D video signals are encoded. Video quality between the left and right views is not always consistent because, for example, each view is encoded at a different bit rate. As a result, the video quality difference between the left and right views degrades the quality of stereoscopic video. However, these characteristics have not been thoroughly studied or modeled. Therefore, it is necessary to better understand how the video quality difference affects stereoscopic video quality and to model the video quality characteristics. To do that, we conducted subjective quality assessments to derive subjective video quality characteristics. The characteristics showed that 3D video quality was affected by the difference in video quality between the left and right views, and that when the difference was small, 3D video quality correlated with the highest 2D video quality of the two views. We modeled these characteristics as a subjective quality metric using a training data set. Finally, we verified the performance of our proposed model by applying it to unknown data sets.
URL: https://global.ieice.org/en_transactions/communications/10.1587/transcom.E96.B.410/_p
Copy
@ARTICLE{e96-b_2_410,
author={Kazuhisa YAMAGISHI, Taichi KAWANO, Takanori HAYASHI, Jiro KATTO, },
journal={IEICE TRANSACTIONS on Communications},
title={Subjective Quality Metric for 3D Video Services},
year={2013},
volume={E96-B},
number={2},
pages={410-418},
abstract={Three-dimensional (3D) video service is expected to be introduced as a next-generation television service. Stereoscopic video is composed of two 2D video signals for the left and right views, and these 2D video signals are encoded. Video quality between the left and right views is not always consistent because, for example, each view is encoded at a different bit rate. As a result, the video quality difference between the left and right views degrades the quality of stereoscopic video. However, these characteristics have not been thoroughly studied or modeled. Therefore, it is necessary to better understand how the video quality difference affects stereoscopic video quality and to model the video quality characteristics. To do that, we conducted subjective quality assessments to derive subjective video quality characteristics. The characteristics showed that 3D video quality was affected by the difference in video quality between the left and right views, and that when the difference was small, 3D video quality correlated with the highest 2D video quality of the two views. We modeled these characteristics as a subjective quality metric using a training data set. Finally, we verified the performance of our proposed model by applying it to unknown data sets.},
keywords={},
doi={10.1587/transcom.E96.B.410},
ISSN={1745-1345},
month={February},}
Copy
TY - JOUR
TI - Subjective Quality Metric for 3D Video Services
T2 - IEICE TRANSACTIONS on Communications
SP - 410
EP - 418
AU - Kazuhisa YAMAGISHI
AU - Taichi KAWANO
AU - Takanori HAYASHI
AU - Jiro KATTO
PY - 2013
DO - 10.1587/transcom.E96.B.410
JO - IEICE TRANSACTIONS on Communications
SN - 1745-1345
VL - E96-B
IS - 2
JA - IEICE TRANSACTIONS on Communications
Y1 - February 2013
AB - Three-dimensional (3D) video service is expected to be introduced as a next-generation television service. Stereoscopic video is composed of two 2D video signals for the left and right views, and these 2D video signals are encoded. Video quality between the left and right views is not always consistent because, for example, each view is encoded at a different bit rate. As a result, the video quality difference between the left and right views degrades the quality of stereoscopic video. However, these characteristics have not been thoroughly studied or modeled. Therefore, it is necessary to better understand how the video quality difference affects stereoscopic video quality and to model the video quality characteristics. To do that, we conducted subjective quality assessments to derive subjective video quality characteristics. The characteristics showed that 3D video quality was affected by the difference in video quality between the left and right views, and that when the difference was small, 3D video quality correlated with the highest 2D video quality of the two views. We modeled these characteristics as a subjective quality metric using a training data set. Finally, we verified the performance of our proposed model by applying it to unknown data sets.
ER -