Projects for NSERC USRA 2012

Department of Computer Science and Engineering
York University, Toronto, Ontario, Canada

Please check back this page regularly as more projects will be added.

Professor Robert Allison

Night vision system development and evaluation for forest fire suppression

The student will develop a prototype simulator for airborne fire fighting based on an open source flight simulator.

Algorithms for image scaling in stereoscopic film

Scaling of stereoscopic content is important for comfortable viewing experiences. This project will investigate automatic processes to accomplish this goal.

Development of software for evaluating self-motion perception in microgravity

The student will assist in the development of software for an experiment aboard the International Space Station.

Visually guided locomotion in virtual reality

The student will develop software to study the effectiveness of locomotion metaphors in virtual environments.

Human perception and performance with 3D displays

The student will assist in the development of user studies for mobile, desktop and large format stereoscopic display systems.

Professors Michael Jenkin and Robert Allison

Motion Platform Interface Design and Development

The MOOG motion platform is Stewart Platform capable of providing compelling displays of real motion. This robotic device is coupled with a stereo head mounted display and appropriate tracking technology to enable users to experience a virtual environment with both visual and physical motion cues. The student will develop a software infrastructure to move the platform safely and precisely for virtual environment simulations and studies of motion perception.

Experience in motion control and robotics is ideal as is an appreciation of software engineering principles for reliability and safety. Students should have a good understanding of C programming.

Professor Amir Asif

Time Reversal Array Imaging Algorithms: Application to Breast Cancer Detection

Breast cancer is the second leading cause of cancer death after lung cancer among women. For years, X-ray mammography was the only practical procedure for detection of breast cancer. However, diagnostic mammography frequently generates many abnormal findings leading to additional, costly imaging procedures and biopsies. This is especially true for young women, who present a higher ratio of dense to fatty tissues in their breast, limiting the effectiveness of X-ray mammography in such cases. More recently, backscatter imaging techniques (based on the seminal work by Fear and Stoica) using low-intensity electromagnetic radiation have been introduced. Malignant breast tumours have electrical properties that are significantly different from those of healthy breast tissues. For example, the cancerous tumour produces a stronger backscattered electromagnetic energy return as compared to the reflections from the normal tissues. However, unlike X-rays, which are non-diffractive and travel in straight lines, electromagnetic microwave propagation in breast tissues is characterized by refraction and multipath effects, i.e., the backscattered cancer signature signal reaches the detector by two or more paths. As a result, standard signal processing algorithms do not perform well due to multipath propagation and cannot accurately identify nor locate cancer tumours with sufficiently fine resolution.

The project will implement a backscatter imaging paradigm based on time reversal signal processing that uses multipath propagation to its advantage for breast cancer detection. The time reversal beamforming imager will be programmed and applied for detecting and locating early stage breast cancer tumours from MRI data.

Professor Melanie Baljko

Dynamic Visualization of Multiple Sensor-Based Data Flows (Digital Media)

This project will entail working closely with a inter-disciplinary team which is developing a large-scale interactive digital media installation that includes multiple networked sensors and feedback loops with human interactants. This project will incorporate elements from Computer Networks (Vlajic, CSE), Human-Computer Interaction (Baljko, CSE), and Spatialized Digital Media (Longford, Dept of Design, Faculty of Fine Arts). It is a joint project between members of Computer Science and the Mobile Media Lab (http://mobilemedialab.ca/). The incumbent will undertake a project to analyze and to redeploy a set of digital media technologies (small objects that will be embedded in natural settings and which sense and transmit environmental and human data); to develop software that will manipulate the data streams that are generated; and to develop different visualization modes for the data. The specification for the visualizations will be developed by the project team. The project report will also contain design recommendations for the design of sensors for the next phase of the project.

Prerequisites: CSE3451 (Signals and Systems); Computer Engineering students preferred

Professor Suprakash Data

Localization in wireless sensor networks

Prerequisites: CSE 3213 is desirable, but not essential.

Brief description: Wireless sensor networks consist of low-power autonomous nodes that communicate with each other using onboard wireless transceivers. Typically, only a small fraction of sensors are equipped with hardware (e.g. GPS) for computing their locations. The other nodes must compute their locations by communicating with other nodes. This is the localization problem.

In this project we will build on previous work on localization to design more robust algorithms for localization, when nodes are mobile. The student will modify existing algorithms an implement them in Java and/or MatLab. Then (s)he will compare the performance of this algorithm with existing algorithms and report the findings.

Background information on wireless sensor networks can be found on the project supervisor's webpage.

Genomic signal processing applications

Prerequisites: Knowledge of MatLab or R desirable. Some background in Molecular Biology or Genomics is helpful but not essential for this project.

Brief description: Computational Biology is an exciting area where computational techniques are used to study real Biological problems. Genomic signal processing involves the use of signal processing techniques to study genomic sequences. It has proven to be very useful in extracting information from Biological data that is of use in understanding diseases and design treatments.

In this project we will investigate the use of Genomic Signal Processing techniques in solving problems in Computational Biology. There are several possible problems that we can consider, including genome identification from a small sequence, microarray probe design, finding repeated elements in DNA sequences and locating CpG islands. Please see the supervisor for more information on any of these problems.

The student will work with the supervisors in designing an algorithm for solving the chosen problem. (S)he will then implement the algorithm (preferably in MatLab or R) and test its performance on real DNA sequences.

Professor Patrick Dymond

Complexity Theory - Alternations and Speedups

Using alternation between existential and universal states, an alternating Turing machine can accelerate a sequential computation. This project will investigate this phenomenon, and apply it in other contexts.
Reference: Ryan Williams, Alternation-Trading Proofs, Linear Programming, and Lower Bounds.

Requirement: Good understanding of complexity theory through courses including CSE 4115 and 3101.

Professor James Elder

Three-Dimensional Context from Linear Perspective for Video Surveillance Systems

To provide visual surveillance over a large environment, many surveillance cameras are typically deployed at widely dispersed locations. Making sense of activities within the monitored space requires security personnel to map multiple events observed on two-dimensional security monitors to the three-dimensional scene under surveillance. The cognitive load entailed rises quickly as the number of cameras, complexity of the scene and amount of traffic increases.

This problem can be addressed by automatically pre-mapping two-dimensional surveillance video data into three-dimensional coordinates. Rendering the data directly in three dimensions can potentially lighten the cognitive load of security personnel and make human activities more immediately interpretable.

Mapping surveillance video to three-dimensional coordinates requires construction of a virtual model of the three-dimensional scene. Such a model could be obtained by survey (e.g., using LIDAR), but the cost and time required for each site would severely limit deployment. Wide-baseline uncalibrated stereo methods are developing and have potential utility, but require careful sensor placement, and the difficulty of the correspondence problem limits reliability.

This project will investigate a monocular method for inferring three-dimensional context for video surveillance. The method will make use of the fact that most urban scenes obey the so-called .Manhattan-world. assumption, viz., a large proportion of the major surfaces in the scene are rectangles aligned with a three-dimensional Cartesian grid (Coughlan & Yuille, 2003). This regularity provides strong linear perspective cues that can potentially be used to automatically infer three-dimensional models of the major surfaces in the scene (up to a scale factor). These models can then be used to construct a virtual environment in which to render models of human activities in the scene.

Although the Manhattan world assumption provides powerful constraints, there are many technical challenges that must be overcome before a working prototype can be demonstrated. The prototype requires six stages of processing: 1) The major lines in each video frame are detected. 2) These lines are grouped into quadrilaterals projecting from the major surface rectangles of the scene. 3) The geometry of linear perspective and the Manhattan world constraint are exploited to estimate the three-dimensional attitude of the rectangles from which these quadrilaterals project. 4) Trihedral junctions are used to infer three-dimensional surface contact and ordinal depth relationships between these surfaces. 5) The estimated surfaces are rendered in three-dimensions. 6) Human activities are tracked and rendered within this virtual three-dimensional world.

The student will work closely with graduate students and postdoctoral fellows at York University, as well as researchers at other institutions involved in the project. The student will develop skills in using MATLAB, a very useful mathematical programming environment, and develop an understanding of basic topics in image processing and vision.

For more information on the laboratory: www.elderlab.yorku.ca

Requirement: Good facility with applied mathematics.

Estimating Pedestrian and Vehicle Flows from Surveillance Video

Facilities planning at both city (e.g., Toronto) and institutional (e.g., York University) scales requires accurate data on the flow of people and vehicles throughout the environment. Acquiring these data can require the costly deployment of specialized equipment and people, and this effort must be renewed at regular intervals for the data to be relevant.

The density of permanent urban video surveillance camera installations has increased dramatically over the last several years. These systems provide a potential source of low-cost data from which flows can be estimated for planning purposes.

This project will explore the use of computer vision algorithms for the automatic estimation of pedestrian and vehicle flows from video surveillance data. The ultimate goal is to provide planners with accurate, continuous, up-to-date information on facility usage to help guide planning.

For more information on the laboratory: www.elderlab.yorku.ca

Requirements: Good facility with applied mathematics.

Low-Cost Three-Dimensional Face Scanning System

Low-cost three-dimensional face-scanning systems have a large range of potential applications in security and retail markets. Our laboratory at York University has recently developed a prototype face-scanning system that has the potential for very low-cost mass production. This project involves the development of a second-stage prototype that is one-step closer to commercialization.

The project will involve systems design and development of a specialized real-time 3D face scanner. A combination of hardware and software design will be required. The student will work closely with graduate students and postdoctoral fellows at York University, as well as researchers at other institutions involved in the project. The student will develop skills in both hardware and software design, as well as computer-vision techniques.

For more information on the laboratory: www.elderlab.yorku.ca

Requirements: Interest in both hardware and software design at the systems level.

Professor Michael Jenkin

Topological mapping of indoor environments

Topological mapping of indoor environments. Autonomous construction of maps of 3d environment is a key problem in autonomous systems. This project will involve building local environmental maps of indoor environments. Students will use video and laser data to classify the local environment as either a hallway or a node and then work with a graduate student to build a mapping algorithm for a large-scale environment.

Knowledge of C desirable.

One-time-pad encryption/decryption algorithms for hand-held devices

One-time-pads provide a cryptographically secure system. This project will involve implementing this system on an ipod/iphone device.

Interest in learning Objective C is desirable.

Underwater tablet window manager

The AQUA robot tablet runs a standard Linux software system but provides only 12 buttons for input. This project will involve the development and testing of a X window manager that supports this limited input capability.

Bomb disposal robot operator simulator

This project involves building a bomb disposal operator simulation using a standard game engine.

Construction of a robotic sailboat

Assist in the development of a robotic sailboat. Development of control, planning and communications systems for the device.

Interest in model building an asset.

Professor Burton Ma

Automated Medical Image Segmentation

Segmentation of medical images is a fundamental step in several kinds of computer-aided surgery. Many automated approaches to segmentation are based on warping a statistical atlas of anatomy to a medical image, but these approaches fail when the image is of deformed anatomy. This research project will examine the use of atlas-free state-of-the-art vision algorithms for automated segmentation of medical images.

Professor Eric Ruppert

To be posted.

Good background in theoretical computer science courses is required.

Professor Wolfgang Stuerzlinger

Better Layout Mechanisms for User Interfaces Toolkits

The layout mechanisms for many GUI toolkits are hard to understand and often difficult to use. This project investigates new, easy-to-understand layout mechanisms and evaluates an implementation of one of them in a comparative user study.

Required skills: Familiarity with any modern GUI toolkit.

Improved Suggestions for 2D Drawing Programs

Displaying suggestions during the drawing process can make it easier for the user to create common and desirable configurations. This project investigates novel user interface methods to make such a suggestive user interface easier to use.

Required skills: basic 2D computer graphics and interactive systems programming experience.

Predicting Visibility Obstructions for a 6DOF Tracker

Previous work of the supervisor generated a novel and highly accurate Virtual Reality tracking system based on (eye-safe) laser diodes, called Hedgehog. The speed and accuracy of this system matches or exceeds the specification of all competing commercial systems. However, this system works only in 5 or 6-sided immersive display environment. Recent work has adapted the system to track successfully with as little as one large display wall and has also characterized the accuracy more comprehensively. The next step is to generalize the system to work in normal rooms, which may have lamps, shelves, etc. Here visibility obstructions play a significant role and predicting these obstructions can be used to increase the tracking accuracy. This project will simulate a new method for visibility obstruction prediction to detect how big an increase in accuracy can be expected.

Required skills: Matrix algebra

Extensions to an Interactive 3D Desktop Virtual Environment

Within my research group, we have developed several systems that enable the user to manipulate 3D geometry with a mouse. User studies show that the new techniques permit much faster manipulations compared to the state of the art implemented in various systems, such as level editors in 3D games (e.g. Halo Forge), modern Computer-Aided Design (e.g. Sketchup), and geometry editors in collaborative Virtual Environments (e.g. Second Life). This project extends the newest prototype by adding the capability to draw lines, rectangles, and other 2D entities onto the 3D surfaces and using these to quickly generate new 3D geometry via extrusion (based on an existing implementation in another system). Building up on this, a simple, yet new, extension is to enable the creation of new geometry through drawing onto silhouettes.

Required skills: Computer graphics

Linux Driver and Capture Synchronization for a 120Hz Camera System

This project involves writing a Linux kernel driver for a 120Hz Camera System, based on a reverse-engineered protocol. Moreover, the project also involves frame accurate synchronization of the cameras to a (computer controlled) blinking laser through the Linux kernel.

Required skills: Kernel-level programming

Measuring Rooms and Calibrating Large Displays

Many people can benefit from knowing the exact shape of a room or a large wall/display surface. In real-estate, interior design, and construction, manual measurements of the individual dimensions are currently still the standard. However, due to construction tolerances, walls are often not planar, openings not rectangular, etc. This causes many problems, e.g. when fitting furniture, installing floors, installing curtains, etc. Moreover, current floor plans optimize space usage with oblique walls, additional corners, etc. which makes measuring square footage difficult. Large interactive displays suffer from similar problems, as it is hard to project an exact rectangle without an external reference and without such a reference it is hard to match input to output. Even for tiled projection displays, an external reference is necessary to achieve seamless blending. This project investigates a new system that uses a laser range finder mounted on a robotic pan-tilt unit to measure the planarity of surfaces and/or the dimensions of a room. Such a system can then also be used to calibrate large displays. Moreover, the system will use a wide field-of-view camera to reference the measurements to the environment.

Required skills: 3D geometry

Robotic Tangible User Interface for Large Tabletops

Tangible user interfaces provide the user with object that they can touch and use as input devices. One example is the use of (tracked) toy houses to perform a city planning task on a large surface. This project implements a new form of tracking/identification scheme for tangible objects via LED arrays mounted on them. Furthermore, and using robotic components, the tangible objects will have the ability to move around autonomously, which enables important functionalities such as undo and replay.

Required skills: Robotics

Professor Natalija Vlajic

Peer-to-Peer (P2P) Botnet Technology

Botnets - networks of computers that are compromised and controlled by an attacker - have emerged as one of the most pressing security issues facing Internet users today. A relatively new class of Botnets - Peer-to-Peer (P2P) Botnets - is considered to be particularly dangerous, as these botnets are shown to be harder to locate, shutdown, monitor and hijack in comparison to (e.g.) IRC- and HTTP- based botnets.

This research project, depending on the interest and qualifications of the candidate, will examine one or more of the following:
1) techniques for analysis of a P2P-botnet operation and estimation of its size
2) techniques for detection of P2P-botnet presence (i.e. presence of P2P bots) in a LAN domain
3) techniques for botnet infiltration