Summarization, Visualization and Browsing of Massive Video Archives

Blue Shift, LLC proposes research to determine the feasibility of using a novel software approach to automatically extract, summarize and visualize geo-spatial and temporal data of massive video archives. Results displayed at 60 times normal speed enable the video analyst to detect: • Subtle patterns; • Anomalies; • Trends over short and long time periods; and • Variations between patterns at different times of day. The video analyst uses point-and-click access to interactively examine and investigate the target video clip in more detail –spatially (pan and zoom), temporally (fly) or both. In addition, the analyst uses point-and-click access to retrieve from the video archive additional video clips related to the target video clip. The analyst searches for relevant video clips by specifying data on location, time, object classification, object behavior or other derived information. The video analyst controls all parameters relating to classification, time compression, summarization and presentation. For a specific video clip, the software automatically: • Calibrates the ground plane; • Segregates objects and background; • Calculates size and velocities of the objects; • Creates an object data base that includes properties of the object; • Classifies objects according to rules established by the video analyst; and • Renders a presentation based on settings established by the analyst.