Junhwan Kim

Ph. D Student, Computer Science Dept, Cornell University
Address: TBD
Phone: TBD
Fax: TBD

Research Interests

I work in the area of medical image analysis under supervision of Prof. Ramin Zabih. I am interested in applying various computer vision techniques, particularly graph-cuts based techniques that Ramin Zabih and his colleagues have pioneered for last few years. Also I am interested in software engineering and have passed the Chartered Financial Analyst (CFA) Level I exam.

I am extremely fortunate to have an opportunity to collaborate with outstanding people. I currently work with Ramin Zabih, Martin Prince, Hale Erel, Richard Watts, and Yi Wang (now at Univ. of Pittsburgh) at Magnetic Resonance Laboratory, Weill Medical College of Cornell University, and Vladimir Kolmogorov and Amy Gale at Computer Science Dept.


Thesis Projects

Part I
Automatic Selection of Mask and Arterial Phase Images for Temporally-Resolved MR Digital Subtraction Angiography
J. Kim, M. Prince, R. Zabih, J. Bezanson, R. Watts, H. Erel, and Y. Wang.
Magnetic Resonance in Medicine, 48(6):1004-1010, 2002.
International Society for Magnetic Resonance in Medicine
(ISMRM), 10th Scientific Meeting and Exhibition, Honolulu HI, May 2002.
XVI Annual International Workshop on Magnetic Resonance Angiography,
Essen, Germany, Oct 2002.
[Paper] [Abstract (1pg pdf)] [Talk Slide 3.9MB] [Web Page]
Part II
A segmentation algorithm for contrast-enhanced images
J. Kim and R. Zabih
International Conference on Computer Vision
(ICCV), Nice, France, Oct 2003.
[Paper] [Poster Slide]
Part III
Visual correspondence using energy minimization and mutual information
J. Kim, V. Kolmogorov, and R. Zabih
International Conference on Computer Vision
(ICCV), Nice, France, Oct 2003.
[Paper] [Poster Slide]


Miscellaneous Projects

Automatic detection and identification of malignant features in contrast-enhanced MR imaging of the breast
Automatic analysis of early enhancement in 2D Projection MRA as a marker of the chronic injury in diabetes
Factorial Markov Random Fields
J. Kim and R. Zabih
European Conference on Computer Vision 2002. LNCS 2352, pp. 321-334, 2002
[Paper (ps)] [Poster Slide 1MB] [LNCS Web Page]
Jigsaw Image Mosaics
J. Kim and F. Pellacini
Proceedings of ACM SIGGRAPH 2002, To appear
[Paper (pdf)] [Talk Slide] [Web Page]

Unsuccessful Attempts

Autocorrection based on Scale Space Theory and Gabor Transform
Wasted time: 2.5 months
Lesson: Don't use SST unless you are Swedish.
AutoDSA based on Robust Statistics
Wasted time: 3 months
Lesson 1: Robust statistics is not so a fancy tool despite what some computer vision guys advocate.
Lesson 2: KISS
Three-dimensional Polygon Mesh Decomposition using Normalized Cut
International Conference on Pattern Recognition 2002. Accepted
Wasted time: 2 weeks
Lesson: No one is as much interested in your work as you are.
An Effective Colorization Framework using Motion Disparity
International Conference on Image Processing 2002. Rejected
Wasted time: 3 weeks
Lesson: Same as above.


Personal Webpage

Since 15 Apr 2002, Last modified: 21 Jul 2003