Open Video Project Toolkit  ( OVP Tk )

          Requirement for digital media archive:

     Due to advancement and accessibility of digital media technology, we are beginning to accumulate large collection of digital media. There is greater needs for facilities to collect, manage, and publish those media. This is especially true to organizations that holds large amount of digital contents. Open Video Project particularly focuses on this issue with videos, for archiving, managing with relevant meta-data, and retrieving with visual clues.

     Creating right meta-data and user interface is important for effective discovery process. OVP was especially intended to be used in educational/research institute, museum, and etc, that discovery and understanding of context matters instead of watching video for entertainment purpose. Tools currently available for this purpose are prohibitively expensive or not opted for creating their own server to have better control in contents and user interface(UI). Having own local server allows more controlled database of our won domain of dataset.

     Open Video Project(OVP) enables organizations on-line video archive, management, and retrieval without significant technical or financial resource. As the distribution is open source, it also allows technical savvy organizations to create more adapted video archive server.

Further reference:

       Open Video Project   This site is no longer current.
       Open Video Toolkit   This site is no longer current.

          OVP Video Parser (ovsurgen):

     OVP is mainly composed of UI, database, web server components, and video parser. My involvement to OVP is to create parser for the suite to create needed visual components from source video files. Users submit source video files along meta-data. The parser creates different types of still images and videos based on scene change detected with flexible user option to be used in visual retrieval process with UI. Currently, there is no straight tools that performs required tasks, so we have built a tool based on the same API as commonly and widely used tools, like mplayer(mencoder), ffmpegX. This allows flexibility of accepting wide variety of input file and to be upto date with current bleeding edge video codecs.


     Common tasks of parser are to create key frame images, fast scan videos, excerpt video, and animated images for UI, all based on scene change.


Key frames:

jpegs are generated with user defined maximum scene change count and resizing. This is mainly used as thumbnails of each scenes in UI, thus default is jpeg files.


fast scan video:

is generated with user defined speed, video format, and resizing. This is mainly used as fast preview of entire video footage, thus default is flv file.


excerpt video:

is generated with N-seconds of excerpt from the original footage for partial preview.


animated images:

are generated to create motion thumbnails of each scenes in UI.


scene change detection:

is currently based on histogram. More sophisticated methods may be needed in future based on context if especially the context is more of educational in nature.





ovsurgen --help

README for ovsurgen
date: June 12th 2009
version: 0.2

./make will completely rebuild ovsurgen
OVP_Toolkit (this dir), ffmpeg, and gd-* should be in the same dir as
ovsurgen has dependency to ffmpeg and gd


ovsurgen is a CLI tool to parse video files for analyzing/reviewing.
This CLI tool was reated to work as back end engine for
OpenVideoProject video server.  See follwing URL for further info:

     http://www.open-video.org/


ovsurgen expects valid single stream source movie file.  Streams after
the first stream will be ignored.  It should accept variety of
format/codec, but not guaranteed.  mpeg and quicktime (.mov) are known
to work.


ovsurgen expects you have following structure in the same directory as
your source data file

./excerpt	    for flv movie
./fastforward	    for fast forward movie
./keyframes/full
./keyframes/large
./keyframes/medium
./keyframes/small
./keyframes/thumb
./slideshow	    for gif animation


ovsurgen --help will show invocation options as following

/*
Usage: ovsurgen [options] -if in-file [options]

options:

    -help|--help      displays this help
    -verbose
    -if filename      input filename (*required)
    -of filename_stem output filename
                       default is input filename
    -interval         fast forward frame interval count
                       default is 64
    -format           output format (default: mpeg)
    -scene            maximum count for scene change detection

  scale factors are mutually exclsive.  Only the last specified
  Later scale factor in user  argument has higer precidence.

    -scale_ratio      output scale 0.01-1.00
    -scale_x          output x-dimension by pixel count
    -scale_y          output y-dimension by pixel count

*/

An example of invocation will be:

  ovsurgen -if inputfile.mpg -interval 64 -format flv -scene 24 -scale_ratio 1.0

where above options for interval, format, scene, and scale are default if you do
not invoke the tool with those options.  -if (input file name) is required.


./clean will clean all output files only from ovsurgen to above
directories leaving directory structure intact.

// end of README for ovsurgen