Tracking by liaoqinmei

VIEWS: 56 PAGES: 14

									   Tracking



DELTA SmartHouse
              Objectives:
• Location of objects – both animate and
  inanimate
• Identification
• Data collection
• Expandability
                Applications
•   Aid for the disabled
•   Security
•   Lighting, power
•   Personalization
    Existing Technology: Location
•   RFID
•   Video
•   Sonar
•   Infrared
    Existing Technology: Identification
•   Fingerprint
•   Facial recognition
•   Hand geometry
•   Middleware
•   Iris recognition
•   Voice verification
•   Signature verification
             » Source:
               www.biometricgroup.com/reports/public/market_report.html
             Existing Technology:
                         RFID


• Benefits
  – Already Accessible
• Drawbacks
  – Non-Passive
  – Not Adaptable
             Existing Technology:
                       Video Tracking
• Locate and Distinguish
   – Follow Movement
      • Airports
      • Sports
   – Contrast
      • Animals

   – Cameras don’t have to be
     stationary
      • Surveillance


• Collect Data
   – Sociological Study
               Existing Technology:
      Finger Printing and Hand Geometry

• Security:                       •For heightened security, all
   – Military                     require a smart card.
   – Hand Geometry Reader
       • Commercially available
                                  •Secure enough for residential
   – Prisons                      setting without smart card
                                      •Preferred method of
                                      DELTA security team
                         Existing Technology:
                                   Facial Recognition
Facial Features Technology                                  Thermal Recognition
•   The process of facial identification incorporates two   Thermal imaging facial recognition systems
    significant methods: detection and recognition.         are based on the principle that infrared cameras
                                                            can capture unique heat emission patterns from
Detection involves locating the human face within an        a person's face.
    image captured by a video camera and taking that        The heat pattern is called a "thermogram". The
    face and isolating it from the other objects captured   thermogram is converted to a digital format by
    within the image.
                                                            the computer and saved as a template in the
•   Software canvasses the captured images for general
    facial structures such as the eyes and nose, and        recognition system's database. The vascular
    measures and determines the rest of the face.           patterns from the ear and nose are removed.
•   After constructing an image of one's face, the
    software "cuts" away any background details leaving
    the image of one's face in a rectangle frame called a   The advantage of this system is the ability to
    binary mask
                                                            capture and identify facial images in the dark.
                                                             Another advantage of thermal imaging
Recognition is comparing the captured face to other
    faces that have been saved and stored in a              recognition is that it does not recognize subjects
    database. One of the ways this is done involves the     based on the surface of the facial skin. Hence
    use of eigenfeatures or eigenfaces.                     reduces any errors that might occur due to
                                                            scarring, facial hair, or eyeglasses.
                                                            The disadvantage lies in acquiring the hardware
                                                            to implement a system, since the hardware can
                                                            be extremely expensive.
                       Existing Technology:
                                      Iris Recognition
•   Iris Identification Background                              •   Actual iris identification can be broken
     –   The idea behind this method is that the pattern of         down into four fundamental steps.
         blood vessels on the retina could be used to                1. A person stands in front of the iris
         identify an individual because of the iris and
         retinal patterns and the uniqueness of them.                     identification system, generally
     –   In the iris alone, there are over 400 distinguishing             between one and three feet away,
         characteristics, or Degrees of Freedom (DOF),                    while a wide angle camera
         that can be quantified and used to identify an                   calculates the position of their eye
         individual.                                                 2. A second camera zooms in on the
     –   Compared to other technologies of the day the                    eye and takes a black and white
         Iris has six times more distinct identifiable
         features than a fingerprint for example.                         image.
                                                                     3. After the iris system has one's iris in
•   How it Works:
                                                                          focus, it overlays a circular grid
     –   In identifying one's iris, there are two types of
         methods that are used by iris identification                     (zone's of analysis) on the image of
         systems: passive and active.                                     the iris and identifies where areas of
     –   The Active iris system requires the user to move                 light and dark fall.
         back and forth so that the camera can adjust and            4. The purpose of overlaying the grid
         focus in on the user's iris. The active iris system              is so that the iris system can
         method requires that a user be anywhere from
         six to 14 inches away from the camera.                           recognize a pattern within the iris
     –   The Passive iris system is different in that it                  and to generate 'points' within the
         incorporates a series of cameras that locate and                 pattern into an 'eyeprint'. Finally,
         focus on the iris. The passive iris system allows                the captured image or 'eyeprint' is
         the user to be anywhere from one to three feet                   checked against a previously stored
         away from the camera(s). This method provides
         for a much more user-friendly experience                         'reference template' in the database
     Existing Technology:
             Voice Verification
•Benefits:
   –Passive
   –Actively Sampled
•Drawbacks:
   –Accuracy
                            Proposal
  Combine Video Identification and Tracking
                Systems:
• Benefits:
   –   Unobtrusive
   –   Precise                       Research at Duke
   –   Expandable                       •Dr. Carlo Tomasi, Duke
   –   Highly adaptable                 CS Department
   –   Much information available,          •Commercially available
       also room for improvement            camera system


• Challenges:
   –   Camera positioning
   –   3-d positioning
   –   Data collection
   –   Maintaining identification
               This Semester:
       Integrate Recognition and Tracking
                  Technologies
•   Run static Analysis
•   Web Cams
•   Still Photos
•   Project in Hudson basement
•   Future
•   Can be adapted to real-time
         Future Semesters
• Adapt to applications
• Integrate inanimate object recognition
  (furniture)
• Video and RFID tracking of groceries and
  other consumables

								
To top