Detecting and Analyzing Multiple Moving Objects in a Crowd

Technology Marketing SummaryWhile human eyes can unconsciously perceive independent objects in coherent motion patterns such as crowds, automated systems have difficulty detecting and counting independently moving objects. A robust algorithm that works with a video recording apparatus to detect, count, and analyze regions of activity in crowds was developed by computer scientists at ORNL. The algorithm selects regions of coherent motion in both time and space, and then identifies sets of tracks that may contain objects of interest.DescriptionThe technology includes an apparatus for identifying a moving object within a series of video images, an image recording device that stores images in a time series, and an image analysis feature configured to execute a program of instructions to identify the object. The program’s steps include identifying several feature point tracks in which all feature points belonging to the same track are located within a polygon of predefined size. It then calculates a trajectory similarity factor where each similarity is a measure of a maximum distance between two tracks.Benefits•Can automatically locate and count objects in high-density crowds with accuracy
•Overcomes some of the difficulties in detecting individual objects that move in unison
•Requires no complex shape or appearance models to select objectsApplications and Industries• Detecting and counting any type of moving object
• Estimating crowd size for event management and planning
• Support for analyses requiring counts of moving objects, such as in medical analysis
• Security and surveillanceMore InformationAnil M. Cheriyadat. Detecting Multiple Moving Objects in Crowded Environments with Coherent Motion Regions, U.S. Patent Application 12/489,589, filed June 23, 2009.Patents and Patent Applications

ID Number	Title and Abstract	Primary Lab	Date
Patent 8,462,987	Detecting multiple moving objects in crowded environments with coherent motion regions Coherent motion regions extend in time as well as space, enforcing consistency in detected objects over long time periods and making the algorithm robust to noisy or short point tracks. As a result of enforcing the constraint that selected coherent motion regions contain disjoint sets of tracks defined in a three-dimensional space including a time dimension. An algorithm operates directly on raw, unconditioned low-level feature point tracks, and minimizes a global measure of the coherent motion regions. At least one discrete moving object is identified in a time series of video images based on the trajectory similarity factors, which is a measure of a maximum distance between a pair of feature point tracks.	Oak Ridge National Laboratory	06/11/2013 Issued

Technology Status

Technology ID	Development Stage	Availability	Published	Last Updated
UT-B ID 200802135	Development	Available	11/21/2011	10/27/2011