Docstoc

new Left Luggage and Theft

Document Sample
new Left Luggage and Theft Powered By Docstoc
					LEFT LUGGAGE AND THEFT




                          -By
                 Mitesh Gupta
                  Shishir Jain
MOTIVATION
   In recent years the demand on video
    analysis application such as video
    surveillance is growing rapidly. Video
    surveillance is commonly used in security
    systems, but requires more intelligent and
    more robust technical approaches. Such
    systems, used in airports, train stations or
    other public spaces, can bring security to a
    higher level.
OBJECTIVES:


 Detection of Left luggage
 Detection of theft
KEYWORDS

   Left luggage

   Abandonment of luggage by the owner

   Attended and unattended luggage

   Theft
PROPOSED ALGORITHM:
BACKGROUND SUBTRACTION
 Background subtraction is done using Gaussian
 Mixture Model.
HEURISTIC APPROACH                FOR DETECTION OF
OWNER AND LUGGAGE

   Luggage
       Predefined height to width ratio
       Range of height (55,75)pixels
       Range of width (60,85)pixels
   Owner
       Width to height ratio should lie between 0.3 to 0.8
       Minimum height of 120 pixels
       Minimum width of 40 pixels
TRACKING BY MEAN SHIFT ALGORITHM
 Creates color histogram of a blob.
 Color histogram is matched in the subsequent
  images to track the blob.
SAMPLE IMAGE AFTER         DETECTING
LUGGAGE




       Green Circle = 2m and Red Circle = 3m
       Person here is in safe radius
SAMPLE IMAGE AFTER           DETECTING LEFT
LUGGAGE




     Person here is Going into the warning zone an alarm
     is raised
LIMITATION AND PROBLEMS
 After BG subtraction the left luggage becomes
  BG object and is forgotten by GMM model.
 Tracking part fails in case of occlusion.

 Identification of owner: no appropriate method
  found till now.
 Luggage is identified using heuristic approach.

 During theft, if thief is occluded then it is difficult
  to analyze the theft.
THANK YOU

				
DOCUMENT INFO