Computer Science Department News

There will be a Colloquium on September 19 at 10:00 AM by Zhaozheng Yin

MONDAY SEPTEMBER 19
E & C S BUILDING AUDITORIUM (1ST FLOOR)
TIME: 10:00 (DONUTS) 10:10 (TALK)
Speaker: Zhaozheng Yin, Missouri S&T (Rolla)

Title: Large-Scale Cell Tracking in Time-Lapse Microscopy Images for Stem CellEngineering and Discovery

Abstract - Vision-based object detection and tracking techniques play animportant role in a broad range of civilian and military applications such asindustry inspection, surveillance and battlefield awareness. In this talk, Iwill introduce a new tracking system that is able to track dense objectpopulations (hundreds to thousands) and determine their spatiotemporalhistories over extended period of time (days to weeks). This system is beingtested in biomedical research that directs the migration and proliferation ofstem cells to meet research and clinical demands on tissue engineering.

Major challenges to existing tracking techniques, in dealing with stem cells inmicroscopy images, are caused by the special microscopy imaging process and thecomplexity of cell behaviors (shape deformation, object occlusion and cellmitotic events). To address these difficulties, I first establish a scientificunderstanding of phase contrast microscope's optical properties for modelingits image formation process. A linear imaging model is found to well explainthe phase contract imaging system. Using this model, I formulate a quadraticoptimization function with sparseness and smoothness regularizations to restoreartifact-free microscopy images. The removal of artifacts allows for obtaininghigh quality object segmentation by simply thresholding the restored image.Then we apply data association algorithms on segmented object candidates andsolve the problem of object tracking by using linear programming.

Brief Bio - Zhaozheng Yin is an Assistant Professor of Computer Science atMissouri S&T. His research interests spread in the fields of Computer Vision,Pattern Recognition, and Image/Signal Processing. In particular, he has beenworking on robust, efficient single or multi-camera visual informationprocessing for segmenting, detecting, tracking objects, and describing theirbehaviors, with applications on biomedical imaging, surveillance, multimediaand video/image scene understanding. Yin received his BS degree in AutomaticControl from Tsinghua University, MS degree in Electrical and ComputerEngineering from University of Wisconsin-Madison, and Ph.D. in Computer Scienceand Engineering from Penn State University in 2009. He was a postdoctoralfellow in the Robotics Institute at Carnegie Mellon University from06/2009-06/2011. He received the best doctoral spotlight award in CVPR2009 andhe was in the finalist of young scientist award in MICCAI2010.