“If we knew what it was we were doing,

it would not be called research, would it?”


— Albert Einstein (1879-1955)

Over the past five years, I have pursued a number of research projects in the field of image processing and pattern recognition. Most of my research targets automatic detection and classification for Intelligent Transportation Systems (ITS) applications. I am currently continuing my ITS research, as well as actively researching infrastructure security, including the design of novel Supervisory Control and Data Acquisition (SCADA) and distributed control systems capable of withstanding external attacks and major component failures.


One of my first research projects revolved around fall detection using embedded smart cameras. The approach used a modified version of Histograms of Oriented Gradients (HOG) to detect fall events.  An initial revision of the device targeted elder-care facilities and was subsequently patented. I explored classification of fall events using artificial neural-networks, before shifting my focus to pattern recognition for ITS, which ultimately became the topic of my dissertation.

For my dissertation research, I proposed a new way of decomposing complex scenes using a fusion of lightweight algorithms (agents). Multiple agents—each with a reliability of 20-60%—would independently analyze complex scenes and report the result to an `arbiter', which would make a decision by combining the results obtained from individual agent reports.  Instead of using a single computationally expensive algorithm with 90% reliability, systems using the proposed approach were able to achieve comparable (or better) results by utilizing dozens of lightweight, lower-reliability agents. Each agent could be delegated to a separate core or specialized processor, which made these systems achieve a five- or tenfold decrease in computation time. A minor part of this research was published in the dissertation, while most of the work is still ongoing.

Some of the ITS-related research on autonomous scene decomposition received awards, such as the "All-University Doctoral Prize" for Best Dissertation, "Best ITS Student Essay" from the Intelligent Transportation Society of New York (ITS-NY) and the "Applicability of Research to Business and Industry" during the Nunan Research Competition in 2013.

One of my present research directions is the application of multi-agent approaches to scene decomposition in the analysis of thermal and true-color video for Weight-in-Motion (WIM), which helps determine weigh station pull-off violators without the installation of expensive in-ground monitoring technologies (using a network of color and thermal cameras installed on the thruway).


Another major research direction that I am currently pursuing focuses on creating a fault- and attack-tolerant distributed network of Remote Terminal Units (RTU) in a SCADA system, which would be able to mitigate attacks and continue operating during critical component failures.