Current MS Students/Stacy O'Malley/CS590
From CSWiki
Contents |
[edit] Week 1 (06/23/07)
Attended first meeting with Dr. Kang (06/21/07) to discuss possible project topics.
Attended class lecture at which Dr. Sun introduced CS590 and the CS wiki and Jon Beaupre presented the "Cool State" online portal project.
[edit] Week 2 (06/30/07)
Added my page on the CS wiki under Current MS Students page. Reviewed list of platforms and current entries on the Category::Platforms page.
Began a presentation on Perl but later decided against it and began a presentation on OpenCV instead.
Attended class lecture at which Dr. Abbott formally introduced the course; student presentations/demos began this week.
[edit] Week 3 (07/07/07)
Finished my presentation/demo on OpenCV.
Attended class lecture for more student presentations/demos; I presented and demonstrated OpenCV to the class. Media:OpenCV.ppt
[edit] Week 4 (07/14/07)
Worked on connected components analysis test for MS project.
Met with Dr. Kang (07/12/07) to discuss project direction; received three papers to read for future implementation and set milestones.
Read paper "Color Information for Region Segmentation" by Ohta, Kande and Sakai. Began writing a program to convert RGB color features to those discussed in the paper. Also began research Ohlander method of color segmentation discussed in the paper.
[edit] Week 5 (07/21/07)
Worked on program to perform connected components analysis of image with converted color features (Ohta paper).
Also worked on pseudocode for Ohlander method of color segmentation.
[edit] Week 6 (07/28/07)
I was unable to find an implementation of connected components analysis for color images with user-defined distance measure in OpenCV so I began writing my own. There is potential here to contribute to the OpenCV library.
Met with Dr. Kang (07/26/07) and discussed issues faced in my research of low level color image processing. Discussed implementing the pyramid segmentation algorithm and completing my implementation of connected components algorithm.
Briefly introduced what I plan to do for my master's project to class on 07/28/07.
[edit] Week 7 (08/04/07)
Reported progress to Dr. Kang via e-mail on 08/03/07.
I wrote a small program to implement pyramid segmentation using the cvPyrSegmentation function. The function only works with 8-bit images so I had to mess around with scaling in order to try and use it on an image with converted color features. I'm not sure if my scaling is correct though. I used the function cvConvertScale for some testing and I also worked on splitting the channels of the image into separate images in order to better visualize the results of converting the color features of an image to those described in the Ohta paper. All this produced interesting results, if not terribly useful.
I did get my own connected components labeling code working for binary images. I'm working on getting it to work for color images with user-defined distance measure. I think most of my difficulties are due to my rusty C++ programming skills and memory management issues. I'm a little obsessed with getting it to work though so I hope to have it working this weekend.
After I get that working I plan to figure out how to select and work with regions of interest and research the Ohlander method more fully.
I began working on my project prospectus (off-line) as I have to present to class again on the 11th.
[edit] Week 8 (08/11/07)
My code for connected components labeling of color images is now working on small simple images using Euclidean distance measure. More testing is needed on larger more complex images and with different distance measures.
Worked on wiki prospectus for second presentation of project topic to the class.
[edit] Week 9 (08/18/07)
I refined my prospectus a little (off-line) by improving the diagrams and adding more text. Began thinking of the major code pieces and how to visualize them in a diagram.
I wrote out and studied the algorithm for the Ohlander method of segmentation and just out of curiosity I tried to find the original paper by Ohlander online but was unsuccessful. Many papers reference it but I couldn't seem to find the actual one (for free, that is). I had jury duty this week so I had to cancel my meeting with Dr. Kang, but I reported progress to her via e-mail on 08/15/07.
[edit] Week 10 (08/25/07)
I read the "Image Inpainting" paper (a number of times) and began thinking of how to implement the algorithm described. I met with Dr. Kang on 08/23/07 and she gave me two more papers to read: "Soft Scissors: An Interactive Tool for Realtime High Quality Matting" and "Optimized Color Sampling for Robust Matting."
Updated prospectus on the wiki and gave my final presentation to class this week.
[edit] Week 11 (09/01/07)
I named my project and updated the wiki to reflect this and "project."
I mostly studied for my finals this week and gave one cursory read of the Soft Scissors paper.
Attended final class to hear other students final presentations.
