Main
Research
Courses
Links
TA
Other

My Research

I am a Research Assistant working under the supervision of Richard Wildes for the Vision Lab in the department of Computer Science & Engineering at York University. I am also a member of York's Centre for Vision Research and have completed my Masters degree as part of the Lab for Active and Attentive Vision under the supervision of John Tsotsos and co-supervision of Richard Wildes.

Current research

  • GeoRegistration

  • Stereo Vision on Robotic Head

  • GPU Programming of Vision Libraries

  • Completed journal publication Detecting Motion Patterns via Direction Maps with Application to Surveillance

  • Developed and tested implementation of 2nd derivative Gaussian and Hilbert Transform of 2nd Derivative Gaussian 3D Steerable Filters

  • Preliminary work on stereo camera calibration and correspondance

Publications

Refereed Posters

  • K.G. Derpanis and J.M. Gryn, Three-Dimensional Separable Steerable Filters, CVR2005.

Open source projects I have contributed to:

  • VirtualDub (Winddows based Video capture/processing util.) - Developed PVN input Plugin (available here)

  • transcode (Video Conversion) - Developed PVN input/output modules for v1.0

  • Lapack++ (C++ Interface for LAPACK Linear Algebra Package) - Ongoing: developed row-order constructors, fixed minor bugs

  • Coriander (Firewire Camera Software) - Added PVN output code (w/K. Derpanis), Help Debug

Thesis related research

  • I have completed my Masters thesis, entitled Automated Surveillance Using Local Dominant Direction Templates with all revisions as of June 2, 2004 (and received my MSc on October 22, 2004); this thesis proposes a method (which I have implemented) of automatically detecting and classifying motion patterns (potentially of security threats) in live surveillance video. For more information, please e-mail me.

  • A presentation of my more recent work on detecting security threats by using local dominant direction templates.

  • A presentation of my initial work on "Detecting Motion Using Spatiotemporal Analysis" based on the ideas of Wildes & Bergen, 2000; the project can detect the dominant direction of horizontal motion and highlight objects moving in the opposing (or any given) direction. Report and video examples exist, but not available online.

  • PVN (Portable Video aNymap) File format specifications and sample code

Unrelated research

  • A paper I wrote on the "Evolution towards 3G Technologies and Beyond;" a technical overview of cellular phone technologies from initial analog systems to current and future 3G voice/data systems.
HalachaJack