Permalink
Find file
Fetching contributors…
Cannot retrieve contributors at this time
48 lines (35 sloc) 1.58 KB
Video Quality Assessment Tool using SSIM (VQATS).
Written by Kah Keng Tay, kahkeng AT gmail DOT com, 2008.
This tool computes VSSIM*, a similarity score between two videos, when given
their image sequences as input. When comparing any two frames, we average the
RGB SSIM values and use that as the overall video SSIM index for the frames.
Features:
- accomodates frame skips and stalls in input videos.
- performs such alignment automatically and quickly.
- minimize computation to keep things fast.
Credit goes to:
- Zhou Wang for the SSIM metric. http://www.ece.uwaterloo.ca/~z70wang/research/ssim/
- Rabah Mehdi for original SSIM OpenCV code. http://mehdi.rabah.free.fr/
- John-Mark Gurney for fib package. http://resnet.uoregon.edu/~gurney_j/jmpc/fib.html
Setup:
1. Install some pre-requisites (if using Fedora Core):
yum install pkgconfig libpng zlib libjpeg libtiff libjasper
2. Download and install OpenCV. Reference: http://opencvlibrary.sourceforge.net/InstallGuide_Linux
wget http://downloads.sourceforge.net/opencvlibrary/opencv-1.0.0.tar.gz
tar xzvf opencv-1.0.0.tar.gz
cd opencv-1.0.0
PKG_CONFIG=/usr/bin/pkg-config PKG_CONFIG_PATH=/usr/local/lib/pkgconfig/ ./configure --without-ffmpeg --enable-static
make
make install
ldconfig -v
3. Download and install FFmpeg.
svn checkout -r 10489 svn://svn.mplayerhq.hu/ffmpeg/trunk ffmpeg
cd ffmpeg
./configure --enable-gpl --enable-pp
make
make install
4. Build VQATS.
make
5. Update eval.py with paths to ffmpeg and the VQATS executable.
6. Compute VSSIM* using eval.py.
eval.py <refvideo> <testvideo>