Software Development Projects
(1992-2005)
Date: May 22, 2005
Kyungnam
Kim,
Computer Science Department,
This
document presents the software projects which I developed 100% or I had
involved in as a team member. It
also details development languages used, period, contractor, etc. Some projects involve hardware
development. Most software programs
are operated on Microsoft Windows. The list below describes my technical skills
related:
TECHNICAL SKILLS |
|
¡¤
Computer languages and platforms: Visual C/C++, MATLAB, Visual Basic,
Borland Delphi, LISP, Fortran, Pascal, HTML, x86 Assembler, PVM, Unix, Linux,
MS-Windows. ¡¤
Hardware design: pASIC,
QuickLogic, VHDL, Verilog,
Synthesis/Simulation tool, Z80 microprocessor. ¡¤
Image processing, vision, graphics: MATLAB, Mathematica,
IDL, SGRP, OpenGL, several commercial and free toolkits. ¡¤
Multimedia packages: Multimedia Toolbook,
Authorware, Flash, Premiere, Photoshop. |
Contents
1.
UMD-BGS (University of
MarylanD BackGround Subtraction Program)
2.
Perturbation Detection Rate
(PDR) Analysis Program
3.
NASA Crew Performance
Analyzer
4.
Multi-camera
Tracking of Occluded People
5.
Video Quality
Measurement Program
9.
EyeTalk – Multimodal Human-Computer
Interface
11.
Eye Tracking
12.
Parallel Programming
of Vision Algorithms on PVM
13.
CAD Tool for
Block Interlocking
14.
Image Database for
Electricity Facility
15.
Personal
Information Management System – Samsung Electronics
16.
Personal Information
Management System – SunJin Fleet Co.
17.
Database for Water
Supply Facility
19.
Internet Guide for
Dummies
20.
Tour Guide for
Department Visitors
21.
Other Projects
22.
My logo artwork
UMD-BGS,
|
Goal |
This background subtraction (BGS)
program processes input video data captured from a static single camera to generate foreground/background binary
image outputs. Our codebook background subtraction algorithm has
these key functions - (i) fast and compact background modeling by a vector
quantization technique, (ii) temporal filtering allowing foregrounds in the
training period, (iii) a unique color model separating color and brightness
evaluation efficiently, (iv) multiple background layers to handle background
changes. |
|
Start/End
date |
Aug 2001 – Aug 2004 (3 years
development period) |
|
My contribution |
80% of 30,000 lines of codes,
Algorithms |
|
Language(Library) |
Visual C/C++ (MFC, DirectShow,
VisionSDK) |
|
Platform |
MS Windows |
|
Features |
Ÿ Inputs can be live video (from any camera accepted
by DirectShow, i.e., IEEE1394, USB camera, etc) , video files
(*.avi, *.mpg, etc), or a sequence of
image files (*.bmp, *.jpg, *.gif, *.ppm) Ÿ Processing speed:
almost 30 fps w/o post-processing Ÿ Setting of parameters
and post-processing options, Diagnostic displays, Event log, Binary detection
images |
|
Customer/User |
UMD students, many people in
computer vision research/industry communities |
|
Remark |
Ÿ More details and
download at http://www.umiacs.umd.edu/~knkim/UMD-BGS/index.html Ÿ Related publication: K. Kim, T. H. Chalidabhongse, D. Harwood and L. Davis, "Real-time Foreground-Background Segmentation using
Codebook Model", Elsevier Real-Time Imaging 2005. |
Main control program and parameter setting
Examples
Perturbation Detection Rate (PDR)
Analysis Program
|
Goal |
The
program is a comparison platform for detection algorithms. It generates a
text file containing detection rates which can be converted to a PDR graph. A
performance evaluation method using perturbation analysis (called
Perturbation Detection Rate analysis) is proposed. It measures the sensitivity of a
background subtraction algorithm in detecting low contrast targets against
background as a function of contrast. It could be used as an alternative of
ROC. |
|
Start/End
date |
Sep. 2002 – Feb. 2003 (6 months
development period) |
|
My
contribution |
60% of 20,000 lines of codes |
|
Language(Library) |
Visual C/C++ |
|
Platform |
MS Windows |
|
Features |
Comparison of 4 detection
algorithms by adjusting parameters and simulation options |
|
Customer/User |
Internal use only |
|
Remark |
Related publication: K. Kim, T. H.
Chalidabhongse, D. Harwood and L. Davis, "PDR:
Performance Evaluation Method for Foreground-Background Segmentation",
submitted to EURASIP Journal on Applied
Signal Processing. |
Main control program and an example frame
Graph generation
NASA Crew Performance Analyzer
|
Goal |
This project is to track NASA
crew in the space shuttle and to analyze their motions, postures, etc. The
ultimate goal of this project is to find out their cognitive states and to
help designing shuttle interfaces, etc in ergonomics-view. (2001-2002) |
|
Start/End
date |
2001 – 2002 (1 year development
period) |
|
My
contribution |
30% of 20,000 lines of codes |
|
Language(Library) |
Visual C/C++ |
|
Platform |
MS Windows |
|
Features |
|
|
Customer/User |
Foster-Miller Inc. (a NASA
sub-contractor) |
|
Remark |
We used our background
subtraction program for this application |
Main control program and an example frame
Multi-camera Tracking of Occluded
People
|
Goal |
Tracking and segmentation of occluded
people using multiple overlapped views. |
|
Start/End
date |
Nov. 2004 – Jul. 2005, (9 months
development period) |
|
My
contribution |
100% of 5,000 lines of codes |
|
Language(Library) |
MATLAB |
|
Platform |
MS Windows |
|
Features |
Here, an appearance model is
created when an individual is detected without occlusion. The appearance
model is a color density as a function of height since a human's appearance
mostly changes along its height.
We assume that this model is good for each view (even though each
camera's output for the same real-world object is slightly different). Given
the appearance model, the prior location, and occlusion information, we
classify each pixel into the best-matched person class using Bayesian rules
(maximum likelihood). In addition
to appearance, a ground homography is used for resolving the correspondence
of each person between views. |
|
Customer/User |
|
|
Remark |
Publication under preparation |
Example frames
Video Quality Measurement Program
|
Goal |
|
|
Start/End
date |
Nov. 2003 – Jan. 2004, (3 months
development period) |
|
My
contribution |
100% of 2,000 lines of codes |
|
Language(Library) |
MATLAB |
|
Platform |
MS Windows |
|
Features |
Ÿ
Q1 – noise: due to sensor
sensitivity, electronic transmission, illumination fluctuation, camera
vibration, etc Ÿ
Q2 – contrast: affected by
camera optics, resolution, etc Ÿ
Q3 – color information: how well
color values are distributed over the intensity range (entropy). Ÿ Q4 – clipping: pixels brighter or darker than the representation bounds (ex,
[0,255]) are clipped. |
|
Customer/User |
Internal use only |
|
Remark |
Related publication: K. Kim, and L.
Davis, "A Fine-Structure Image/Video Quality
using Local Statistics" IEEE International Conference on Image
Processing (ICIP) 2004 |
|
Goal |
Developed a face animation system. |
|
Start/End
date |
Sep. 2000 – Dec. 2000, (4 months
development period) |
|
My
contribution |
100% of 10,000 lines of codes |
|
Language(Library) |
Visual C/C++ (OpenGL) |
|
Platform |
MS Windows |
|
Features |
It can automatically, in
real-time, animate facial feature movements such as eye-lid blinking, mouth
opening/closing, and head rotation (roll, yaw, pitch) by visual tracking on
the facial features from live video. |
|
Customer/User |
|
|
Remark |
Course Project – ¡°Advanced
Computer Graphics¡± |
Example
Head rotation (roll, yaw,
pitch)
mouth opening/closing
Eyeball movement
|
Goal |
Implemented a face recognition
system based on Eigenfaces papers using MATLAB and tested on the face
database from AT&T Lab at |
|
Start/End
date |
Sep. 2001 – Dec. 2001, (4 months
development period) |
|
My
contribution |
100% of 1,000 lines of codes |