|
Product lead at trg | tech in charge of seeing the product through from concept to completion. Interfacing with cross functional teams, working with designers, developers, marketing, QA etc on a daily basis. Masters in Computer Science from Tufts University and a Bachelors with Honors in Computer Science from Mount Holyoke College. Previously worked at Avid Technologies and Nokia in Boston, Soft Image and Millfield Partnership in Montreal and London respectively. My research interests lie in image processing, physics based animation, data visualisation, next generation interfaces and scientific simulation. Projects I have worked on: |
|
Masters Project: Marble Track Audio Manipulator (MTAM):
A Tangible User Interface for Audio Composition
"We created a tangible user interface that allows children to create musical compositions through constructive play. Our Marble Track Audio Manipulator (MTAM) is an augmented marble tower construction kit where marbles represent sound clips and tracks represent different sound effects. To create musical compositions, children collaboratively build a marble tower and then play their compositions by dropping marbles into the tower. As marbles roll through the tower children can interact with the marbles and thus improvise and alter their musical compositions. By augmenting a popular toy, physically representing sound clips and effects as well as allowing improvisation, the MTAM system provides children with a creative, playful, and engaging encounter with music." Full paper ...
|
|
Medical Imaging at
CIMIT (see link)
The goal of the project was to review the use of level sets, fast marching, and thresholding in Slicer (www.slicer.org/) to improve the efficiency of segmenting the vascular structure within the brain. We specifically focused on the anterior cerebral arteries that emerge from the circle of Willis. The problem was to detect the existence of these two tiny arteries without allowing them to merge into a larger set. The technique used now in the CIMIT Sim Group is leading to the merging of those two arteries and is therefore giving erroneous information on the vascular network at this location. Manual image processing is needed to correct this and obtain two distinct arteries. Setting a new technique to segment the two anterior cerebral arteries separately would improve the segmentation technique and save time. [Back to top] |
|
|
Thesis: Real-Time Interactive Ray tracing in Distributed Environments for
Virtual Reality Systems The broader goal of this paper is to increase the efficiency and realism of rendered images in virtual reality systems while preserving a relatively high frame rate. Ray tracing has a high degree of realism but is very costly in terms of computational expenses. For this reason it is not recommended for use when real time interactions are required in graphics such as in virtual reality environments. This investigation looks at developing an algorithm which works alongside the ray-tracing algorithm and interweaves estimations based on the temporal coherence inherent in certain types of scenes with actual ray traced frames thereby increasing the frame rate while preserving the visual quality of the environment. This algorithm applies to environments, which allow for only one degree of freedom in the users motion i.e. the motion is along the direction of the camera (so the user can only walk looking ahead on a linear path consistent with our experimental treadmill environment). It is also possible to add a rotational component thus allowing this algorithm to be applied to rotational motion as well as linear motion. The algorithm is developed in a way that it is distributable allowing it to be combined with parallel processing power to increase the speed even further. The system will increase the quality, speed and efficiency of interactive ray tracing required to generate the graphical environment for our immersive locomotive virtual environment. This paper will first explain some of the background of ray tracing in distributed environments such as the Beowulf system. Then it will talk about how my algorithm uses geometric constraints in the environment and the constraints on the motion of the user as indicators of the optic flow patterns in the scene. This optic flow will be used to determine how the images change from frame to frame in a computationally less expensive approach, which intends to be qualitatively as effective as actual ray tracing. Complete thesis ...
|
|
Perceptual Aspects of Locomotion Interfaces for Virtual environments
The goal of this research was to understand the perceptual cues required in a virtual environment to allow the user a realistic experience. This involved the user's ability to accurately percieve or judge the depth of objects in a scene as well as a correct assessment of their speed relative to said objects. | To do this we conducted experiments where a user first walks through a real and then a virtual environment and comparing the user's actions, we determined the amount of recalibration required in the virtual world. The virtual world could be represented with different spatial and visual properties (such as textures) thus providing the user with various cues to judge depth and speed. The final goal would then be to minimise the amount of recalibration required in the virtual environment. |
[Back to Top]
|
|