The end product of the analyses described in Part 1 is metadata.   To be sure, metadata is defined as “structured information that describes, explains, locates, or otherwise makes it easier to retrieve, use, or manage an information resource.  Metadata is often called data about data or information about information.” [NISO Press, National Information Standards Organization, 2004, ISBN: 1-880124-62-9 ]

Originally the purpose of creating metadata was to facilitate discovery of relevant information.  However, today with the continued exponential increase of video source, metadata takes on other purpose and that is to synopsize the content in a structured manner so that one can (1) get the gist of the content of a long video sequences (minutes to hours) without spending the time to view it and/or (2) to be able manage large holdings of video data (petabytes) and retrieve portions of interest using queries on the metadata.  This same metadata analogy applies in the audio domain or combined audio/video domain.

The roadmap is intended to be a summary of generic video analysis technologies that cover a broad spectrum of video sources.  The video sources can be generalized into the following categories:

• Broadcast News (of particular interest is foreign news) wherein the program video and audio are captured/recorded
• Surveillance video obtained from stationary or panning cameras placed inside and/or outside a facility to observe moving objects (e.g. people) within the facility, it’s immediate compound, or outside it’s perimeter
• Meeting Room video that captures a video (and generally an accompanying audio record) of a meeting proceedings
• Overhead Aerial Surveillance  video collected of the ground beneath by manned or unmanned vehicles flying overhead  (airplanes or UAVs) in panning and/or zooming modes as the aircraft overflies the terrain from low, medium or high altitude
• Ground Reconnaissance video taken by tourist or any other person who records imagery of buildings, places of interest, scenery or whatever the viewer’s eye sees that is collected when the camera is stationary but panning or when the camera is in motion, e.g. from a moving platform (person walking or automobile).  As cell phones become more ubiquitous and sophisticated they are becoming an ever growing source of video collection

To analyze and extract metadata from these various video sources requires the use of numerous analytic techniques, some common across many collection sources while other algorithms have focused applications in extracting information from specific video sources.