R. Ismail Haritaoglu

Graduate Research Assistant

University of Maryland-College Park

Office: AVW 2132

• Office Phone : (301)- 405 7042
• e-mail : hismail@umiacs.umd.edu

• Resume: [html][text]
• Research Summary:[html][text]
• Publications:

Academic Degrees:

• PhD (in Progress) Computer Science Dept., University of Maryland, College Park

• M.S. (September 1994) Computer Engineering and Information Science Dept. Bilkent University, Ankara, Turkey

• B.S. (June 1992) Computer Engineering and Information Science Dept. Bilkent University, Ankara, Turkey

PhD Research Description:

• W4: A PC based  real-time visual surveillance system for tracking  people and their body parts, and monitoring  their activities in monochromatic andimagery.It operates on grayscale video imagery, or on video imagery from an infrared camera. Unlike many systems for tracking people, our system make no use of color cue. Instead W4 employs a combination of shape analysis, robust tracking techniques, silhouette based body model to locate and track the people  and understand the interaction between people and objects - e.g., people exchanging objects, leaving objects in the scene.

• W4S: A subsequent system, integrated real-time  stereo computation   into  W4. Incorporation of stereo has allowed us to overcome the difficulties that  W4 encountered with sudden illumination changes, shadow and occlusion which makes tracking much harder in intensity images.

• Ghost: A  new silhouette-based body model Ghost  is described to determine the location of body parts while the people are in generic postures. It is a combination of a hierarchical body pose estimation,  a convex hull analysis of the silhouette,  and  a partial mapping from the body parts to the silhouette segments  using a distance transform method that incorporates the topology of the  human body.

• Hydra: A real-time system for tracking multiple people in monochromatic imagery, we describe the computational models employed by Hydra to track multiple people before, during and after occlusion, and where the people can not be isolated. It combine a-silhouette-based shape mod el, motion model, correlation-based matching methods to classify whether or not the foreground blob contains multiple, determine the location of people, track them. Unlike the other previous systems, Hydra can build appearance model and segment them du ring tracking to identify individual person before, during, after occlusion.

"Who is Carrying an Object?" We have designed a silhouette-based method to determine whether or not a person is carrying an object. It combines  two basic observations: Human body shape is symmetric, and people exhibit periodic motion while they are moving. During tracking, the people's periodic motion of people is estimated, and the regions on the silhouette which violate the symmetry-constraints are determined. Those results are combined to determine if a person is carrying and object, and to segment the object from the silhouette.  In cooperation with Hydra during tracking, we construct an appearance model for each carried object, so that when  people exchange  objects, we  can detect "who" carries "which" object via an analysis of the segmentation.

• ActiveTracker:  A system to track single target with active camera (which has pan, tilt ,zoom control) in a large field of regards, When W4 detects people which appear very small in image space or when a person is about to leave the current field of view, W4 can give the control of tracking to ActiveTracker . It zooms into area to get sufficient resolution to recognize human actions, and actively track them using motion cues by controlling the rotation of the camera.

• Adaptive Background :  Statistical background scene modeling utilized in W4 allow us to segment foreground object, even background is not completely stationary,(e.g. motion of tree branches). W4 is capable of learning the background even the scene contains moving people, and updating the background model parameters periodically during tracking. Also W4 keeps the log of tracking, and background models during surveillance, and it includes object, which was detected and then stay motionless for a long time, into background model (e.g. parked cars, deposited objects).

•  Shall We Dance ?: A real-time 3D computer vision system for detecting  and tracking human movement. Multiple cameras observe a person; silhouette analysis and template matching achieve real-time 3D estimation of human posture Dynamics/kinematics model of human body and Kalman filter are utilized to help the tracking process as well as to interpolate some joint locations. All estimation and rendering processes run in real-time on a PC based system.

• Other Research on Computer Vision

Other Research and Projects:

• Real Time Video Teleconferencing
• Image Mosaicing
• Software Regression Testing with The Omega Test
• Pattern History based Branch Prediction Techniques
• Design Automation of FPGA's
• Parallel Processing

Ismail Haritaoglu  23 Observation Ct.. #304  Germantown, MD 20876

(301) 515 09 63

Send me an e-mail!  hismail@umiacs.umd.edu

Last updated on September,25  1996