Research projects - from the most recent




Clutter cover     Proto-object Segmentation 


     The goal for this project is to quantify the number of "things" in

     an image, which we call the proto-objects - image regions that
     have consistent low-level features. Proto-objects are segmented
     using a novel parametric graph partitioning method, and used
     to model human visual clutter perception. The method can also
     be used as a clustering algorithm.


 

tumor cover     Brain Tumors Detection and Segmentation in MRI 


     The precise segmentation of brain tumors from MR images is necessary

     for surgical planning. However, it is a tedious task for the medical professionals
     to process manually. This project is an unsupervised, fully automatic brain
     tumor detection and segmentation method based on utilizing the blob-like
     characteristics, as well as the brain billateral asymmetry that results from the
     presence of tumors.

 

 

Clutter cover     Modeling MST Neuron Receptive Fields


     The neurons of the Medial Superior Temporal (MST) area are

     parts of the "where" pathway of the dorsal stream, in visual
     processing system. They are responsible for the perception of
     self-motion, and we explore ways to model the MST neuron
     receptive fields using a dual-Gaussian models.

 

 

Clutter cover     Deformable Contour Tracking


     This is the final project for the Advanced Computer Vision course

     at Penn State University, Spring 2009, taught by Dr. Robert Colllins.
     We focused on object tracking using deformable contours, and we
     implemented the Hidden Markov Model method by Huang et. al 2006,
     and tested on hand movements as well as brain MRI frames.

 

 

Clutter cover     Cerebral Assessment Systems - Navigation


     A virtual-reality of the Strong Memorial Hospital (URMC) at Rochester, NY.

     It is a system with various different tasks-oriented tests for assessing a
     person's navigation impairment. The virtual-reality was constructed using
     Quake 3 game engine in C, and all the tests were implemented using the
     same engine.