|
Video Quality
Group at Faculty of Electrical Engineering (VQG@ETFOS) |
||
Welcome to the
web page of Video Quality Group at Faculty of Electrical Engineering, University of Osijek, Croatia.
The researchers in this group are:
Video
quality databases
VQG@ETFOS has developed
two different video quality assessment databases in two different resolutions (CIF and VGA). These databases are publicly available to the researchers which would like to evaluate the performance of their quality assessment algorithms. ETFOS
CIF Video Quality (ECVQ) database
The ECVQ database uses eight uncompressed
progressive video sequences with
significantly different content (in terms
of spatial and temporal activity). The reference sequences are in the raw YUV 4:2:0 format, all in
CIF (352x288) resolution with
25 fps (frames per second) and
with the total of 300 frames. A set of 90 distorted videos were created
from these reference videos (the number
of distorted videos for each reference video
is specified in ''Readme'' file for a given database). Two different distortion types are used: H.264/AVC compression and MPEG-4 Visual compression. Sequences are coded in different bit rates from 73 to 827 kb/s, which span a wide range of distortion.
The experimental subjective quality evaluation for 40 non-experienced observers is performed according to Subjective Assessment Method for Video Quality evaluation (SAMVIQ). Every
video sequence, including
the hidden reference sequence, was presented to the observers only once. Observers had to evaluate the video quality by using
the slider with the scale
from 0 to 100. After screening of observes, difference mean opinion scores (DMOS) were calculated on a per subject per test video sequence basis. The mean and
variance of the DMOS obtained from the subjective
evaluations, along with the reference and distorted videos, are available as part of the
database. ETFOS
VGA Video Quality (EVVQ) database
The EVVQ database uses eight uncompressed
progressive video sequences with
significantly different content (in terms
of spatial and temporal activity). The reference sequences are in the raw YUV 4:2:0 format, all in
VGA (640x480) resolution with
25 fps (frames per second) and
with the total of 300 frames. A set of 90 distorted videos were created
from these reference videos (the number
of distorted videos for each reference video
is specified in ''Readme'' file for a given database). Two different distortion types are used: H.264/AVC compression and MPEG-4 Visual compression. Sequences are coded in different bit rates from 261 to 1737 kb/s, which span a wide range of distortion.
The experimental subjective quality evaluation for 40 non-experienced observers is performed according to Subjective Assessment Method for Video Quality evaluation (SAMVIQ). Every
video sequence, including
the hidden reference sequence, was presented to the observers only once. Observers had to evaluate the video quality by using
the slider with the scale
from 0 to 100. After screening of observes, difference mean opinion scores (DMOS) were calculated on a per subject per test video sequence basis. The mean and
variance of the DMOS obtained from the subjective
evaluations, along with the reference and distorted videos, are available as part of the
database. Copyright Notice -----------COPYRIGHT NOTICE STARTS WITH THIS LINE------------ The following two papers are to be cited in the bibliography whenever one of databases is used as: ·
S. Rimac-Drlje, M. Vranješ, D. Žagar, Foveated mean squared
error – a novel video quality metric, Multimedia
Tools and Applications, vol. 49, no. 3, 2010, pp. 425-445 ·
M. Vranješ, S. Rimac-Drlje, K. Grgić, Review of Objective Video Quality Metrics and
Performance Comparison Using Different Databases, Signal Processing-Image
Communication, 2012, http://dx.doi.org/10.1016/j.image.2012.10.003 The reference videos in this database
are the original copyright-free videos that
were available from the Video Quality Experts Group. We processed the reference videos using various techniques to obtain the distorted
videos in this database. IN NO EVENT SHALL THE UNIVERSITY OF OSIJEK, FACULTY OF
ELECTRICAL ENGINEERING BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL,
INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OF THIS DATABASE
AND ITS DOCUMENTATION, EVEN IF THE UNIVERSITY OF OSIJEK, FACULTY OF
ELECTRICAL ENGINEERING HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -----------COPYRIGHT NOTICE ENDS WITH THIS LINE------------ Download
The ECVQ and EVVQ video quality databases are freely available to the research community. If you use some of these databases in your research,
we kindly ask that you
reference our next two papers:
The information necessary for download of databases can be obtained
by contacting Mario
Vranješ (mario.vranjes@etfos.hr).
Please indicate your university/industry affiliation and a brief description
of how you plan to use the database(s). ECVQ database can be downloaded
here, while EVVQ database can be downloaded here. |
||