Crowd state characterization system and method

摘要

A crowd state characterization system utilizes a plurality of processors to analyze video streams from numerous videos to select videos and/or video frames of interest. The processors digitize dismounts such as pedestrians and the like and then analyze the digitized pedestrians. The frames of video are characterized in terms of entropy related to discordant motion and enthalpy related to energy. A selector can then select from among numerous videos to allow observation of videos numerically determined to be of interest.

主权项

1. A system for automatically characterizing a plurality of video streams, comprising:
a plurality of cameras that produce said plurality of video streams, each video stream comprising frames of video; and a plurality of processors configured to receive said plurality of video streams, each of said plurality of processors being programmed for digitizing a plurality of dismounts within said plurality of video feeds, said plurality of processors being programmed for producing numerical characterizations for said plurality of dismounts in said frames of video of an enthalpy related to an energy of said plurality of dismounts and an entropy related to a discordance of motion of said plurality of dismounts with respect to a comparison of said frames of video, whereby each of said video streams is numerically characterized for said enthalpy and said entropy; wherein said plurality of processors are programmed to determine crowd region of influence related to a density of said dismounts, and to produce a numerical value for a crowd temperature based on a size of said region of influence, a number of said dismounts and said enthalpy and said entropy.