Part 1:  Introduction
Part 2:  Video Basics
Part 3:  Roadmap Description
Part 4: Requirements

Independent of requirements, the need for autonomously dissecting a video sequence has many technology drivers.  In addition to the ubiquity of video cameras that make video a simple and logical recording source, the emergence of commercially available low-cost, high-definition cameras, massive storage capabilities and phenomenal computer speed combine to make video the recording media of choice.  With respect to extracted metadata video provides so much more information theory entropy than a photograph. For example, a photograph enables one to develop a portrayal of a scene or activity as a sequence of nouns and adjectives while a video, because of its temporal dimension, enables the derivation of additional metadata in the form of verbs.  So what do all the technological drivers mean?  High-definition video cameras enable the collection of higher resolution data, which, in theory, enable the capture of more information thus making extraction of greater detail possible.  While providing this additional information potential the negative side of HD video is that there are so many more content bits to transmit, store and analyze.  Great strides have simultaneously been made in storage capacity as well as the reduced cost per unit storage.  Magnificent increases have been and continue to be realized in the innate computer power available for processing this enormous amount of data. This capability is being realized by virtue of highly parallel machine structures and more generalization of graphic processing chips available or applied in a cloud computing scenario.

As described in the requirements (Part 4), the key ingredients to improving the quality of autonomous computer processing of is improving analytic algorithms to take advantage of the ever increasing entropy.   So technology is a double edge sword – the supply side of technology continues to produce new, powerful, elegant and robust collection sources that produce an inordinate amount of data containing a wealth of information while straining the processing side to keep up with the amounts of data to be handled, stored and managed, to enable processing the resultant data in an efficient manner (typically real time or near real time) and to a level of confidence (precision, recall) that a military commander has the necessary assurance that he has the “picture right” before he pulls the trigger.