ECE 538 Image Analysis & Computer Vision II - 3 hours credit


Professor: Dr. Scott E Umbaugh Office: Engineering Building, Room EB3037

Phone: 650-2948 e-mail: sumbaug@siue.edu

Class Format: Seminars and project. A series of seminars, with topics of current interest in computer vision application research related to the project, will be presented by the professor. The students will also participate by presenting their own project.

Description: Topics of current interest in computer vision. Applications of pattern recognition, image analysis, multi-spectral computer vision. Group projects.

Objectives: To familiarize the student with current areas of research in computer vision. Various presentations by the professor and students will be used to achieve this goal. The students will become familiar with the literature -- journals, magazines, conferences, etc. -- in this research area.

Prerequisite: ECE438, or consent of instructor

Reference Texts: Pattern Recognition, 4th Edition,  S. Theodoridis and K. Koutroumbas, Academic Press, 2009,  Digital Image Processing and Analysis: Human and Computer Vision Applications with CVIPtools, 2nd Edition, SE Umbaugh, CRC Press, 2011

 

Course Schedule:

Week

  • 1....... Meetings with professor for project discussion, Pattern Classification Seminars, Image Fidelity and Presentation Seminars
  •             Reading: Umbaugh – pp. 104-125, 335-387; Theodoridis & Koutroumbas Intro & Chaps related to project
  •             Homework: Umbaugh book Chapter 6: Exercises: 1-30, Supplementary Exercises: 1-10
  • NOTE: You can use your own homework on the Quiz next week
  • 2....... Pattern classification quiz, Meetings with professor for project discussion
  • 3....... Journal paper/project proposal presentations, project meetings
  • 4,5... Seminars/project meetings
  • 6...... Progress presentations by students, project meetings
  • 7...... Seminars, project meetings
  • 8...... Project presentations

GRADE:

  • 20% Pattern Classification Quiz
  • 10% Project proposal and journal paper presentation
  • 5% Progress presentation
  • 65% term project – paper (see below), presentation

 

TERM PROJECT

This semester we will focus on developing computer vision research/application projects selected by the student, subject to approval. Specifically we will be working in two research areas: 1) Skin lesion/dermatological image analysis, and 2) veterinary thermographic image analysis. Students are encouraged to work in groups of 2 for the projects. Groups of 3 or more will NOT be allowed (don’t ask).

1. Skin Lesion Detection and Evaluation

2. Veterinary Thermographic Image Analysis

During the term you are required to submit three evaluations of the work performed by each member in your group, including yourself. These evaluations are as follows:

  • These are all confidential, the only person to see them is the Professor. The Professor will make final grade decisions.
  • Justify the grades you assign with specifics - for example, "we scheduled three meetings, student X always showed up prepared, or student Y was never on time and did not have their part of the project completed".
  • These will be emailed directly to me at sumbaug@siue.edu before each of the three milestones – proposal, progress report presentation and final report.
  • Include your name, group members names, which of the three milestones and the date. A short evaluation should be written about each member of your group, including yourself.
  • Two items for each person: 1) Your evaluation of their work in words. 2) A number of points based on the following:
  • 5 points are to be allotted for each person, e.g., 2 people give a total of 10 points. If you feel you both contributed equally, give 5 points to each. If you feel you did a little more, give yourself 6 and the other person 4. In other words, distribute all the points according to the amount of work each person contributed to the project. Note that this is a zero sum process - the total must add up to 10. These evaluations will be used as part of your grade, and will be used to determine individual project grades.

A paper will be written describing the project and discussing what was learned during the project. In addition to the paper each group will do a presentation for the class describing their project. The final paper will be about 25 to 50 pages, typed, double-spaced (excluding appendices). Include images in the paper! In addition to the paper each group will do a presentation for the class describing their project. The following number of pages is a suggestion and an approximation, not an absolute!

Ø  In addition to handing in a paper copy of the report, email me a soft copy of the Word file. Before you send me the file give it a meaningful name that includes your last name(s) and the project title.

Paper Format Outline

  • 1. Title page (title, names, course number, date, etc.)
  • 2. Table of contents with page numbers for: different sections, figures, appendices, etc.
  • 3. Abstract - 1 page or less. Concise description of what is contained in the paper, include brief summary of results.
  • 4. Introduction/Project overview - 1 page.
  • 5. Body of paper. Broken down into sections as required for your part of the project. For example: Background/theory, experimental methods, discussion and analysis of results, program descriptions, etc. Present results using graphs, images, etc., 10 to 25 pages
  • 6. Summary and conclusions. Summarize any results and draw conclusions as based on these results. 1 to 4 pages.
  • 7. Suggestions for future work. Include any ideas you have based on your work and conclusions about followup experiments and/or research. 1 to 2 pages.
  • 8. References. Be sure your references are complete. Avoid web sites as references – these come and go – find the source, which is usually a published paper.
  • 9. Appendices - related background information, program listings, etc.

General: reports should be typed, double spaced, pages numbered starting with abstract. The number of pages listed above are only guidelines, do what is necessary, but keep it concise. DO NOT put in plastic folder, simply staple in upper left hand corner.

The students will give a presentation of the project during the last week of the semester.

Grading: The project is worth 65% of your grade, broken down as follows:

  • 15% Difficulty and complexity
  • 20% Quality of work and success
  • 10% Quantity of work
  • 10% Project paper
  • 10% Project presentation

 

COMPUTER RESOURCES AVAILABLE

Hardware:

  • 19 Windows imaging workstations, frame grabber and image compression boards
  • HP color scanner, 1200 dpi
  • HP Color Laser Printer
  • Digitizing stations with CCD cameras, zoom and standard lenses, controlled light boxes
  • Sony digital Mavica still/MPEG camera, XGA resolution (768x1024)
  • Canesta’s DP205 3-D camera

Software:

  • CVIPtools: a comprehensive Computer Vision and Image Processing package developed at SIUE – Windows and UNIX
  • PARTEK: Pattern Recognition software with tutorials, comprehensive statistical pattern recognition and some neural network functions - Windows
  • Image Alchemy: comprehensive image conversion and compression package - UNIX
  • Khoros: comprehensive CVIP development environment - UNIX and X-windows
  • xv: on-screen image manipulation - UNIX/X-windows
  • Matlab: Image Processing, Neural Network, and Digital Signal Processing Toolboxes
  • Image databases: Image Databases
  • Robot project image database: 438project.html
  • Microsoft Office, word processing, presentations, etc
  • Microsoft Visual Studio 6.0

THE RESEARCH ENGINEER'S NOTEBOOK

NOTE: In ECE 538 you are required to keep a research engineer's notebook which will be reviewed by the professor during meetings.

INTRODUCTION: The technical notebook is one of the most important tools for any engineering work. This includes: basic research, product development, or engineering design. It is primarily for the researcher's own use, but another person with similar technical background should be able to understand and duplicate any experiment, data, and conclusion, or to prepare a technical report

following only the notebook.

There are many reasons to keep an accurate and complete record of your work:

  • 1) to establish the authenticity of the work.
  • 2) to defend patents.
  • 3) to act as a basis for technical reports and articles.
  • 4) to avoid duplication of effort.

The nature of the work and the purpose of the research will influence the content and format of the notebook.

CONTENT REQUIREMENTS: The notebook must be understandable to aperson with a comparable technical background. It must be legible. It must be complete; for example, "We got code from book" is NOT an acceptable entry - what code ?, what page ?, what does it do ?, did you have to recompile it ?, etc.

The notebook must answer the following questions:

  • WHAT WAS DONE? This includes the approach to the research problem. Any ideas generated should be included. Algorithmic flowcharts, references used, notes taken, etc. should be included.
  • WHO DID IT? List all those who participate in the project for a given entry, including yourself, at the beginning of each entry. Any corrections or alterations should be initialed.
  • WHEN WAS IT DONE? It must be obvious to any reader when the work was performed. Date all pages and entries; entires that extend beyond one page should be dated on each page. Do not leave blank spaces and NEVER "back-date" entries (NEVER make ANY false entries in your engineering notebook).

General: The typical engineers notebook available in bookstores will be blue, brown or black, is approximately 9" X 12", and has about 100 to 150 pages. The notebook will be bound, never looseleaf, and the pages should be numbered consecutively, preferably by the printer. For the our purposes you may use spiral notebooks, as long as each page is numbered and each entry is dated.

A neat, organized and complete notebook record is as important as the investigation itself. The notebook is the original record of what was done. It is not a report to be written after completing an investigation. Do not write on scratch paper expecting to transfer it later to the notebook. Use a blue or black non-eraseable pen. Errors are not erased, but simply marked through with a single line so that they still can be read - later you may discover that your "error" contains important information.

Leave the first page or two in the notebook blank for a Table of Contents. This is necessary so that your work can easily be referenced. Use only the right-hand, odd-numbered pages for the notebook record. Use the left-hand, even-numbered pages for sketches, rough calculations, and memos to yourself. You may also place diagrams and graphs on the left, opposite corresponding procedures and calculations. Do not leave any blank spaces/pages in the notebook.

Format - Technical Diary

Organization of this format type is left to the engineer. This format is suited to experimental work, design work, and research. The general format and content requirements must be met. Notes, program code, flowcharts, procedures, data, and calculations are blended together logically and chronologically to form a step-by- step diary describing work. Observations and conclusions are entered as they are made, and summarized at the logical end of a section. This format is well suited for research.


Brief Bibliography

Books

  • Digital Image Processing and Analysis: Human and Computer Vision Applications with CVIPtools, 2nd Edition,  Scott E Umbaugh, CRC Press, Taylor & Francis Group, Boca Raton, FL, 2011, 956 pages, ISBN: 9781439802052
  • Computer Imaging: Digital Image Analyis and Processing , Scott E Umbaugh, The CRC Press, Boca Raton, FL, January 2005
  • Digital Image Processing, R.C.Gonzalez & R.Woods, Addison-Wesley, 2002
  • 1a.Computer Vision and Image Processing, S. E Umbaugh, Prentice Hall, 1998
  • 1. Computer and Robot Vision, R.M. Haralick and L.G. Shapiro, Addison-Wesley, 1992
  • 2. Machine Vision, Jain, Kasturi, Schnuck, McGraw-Hill, 1995
  • 3. Robot Vision, B.K.P.Horn, MIT Press, 1986
  • 4. Computer Vision, D.H.Ballard & C.M.Brown, Prentice Hall, 1982
  • 5. Object Recognition by Computer, W.E.L. Grimson, MIT Press, 1990
  • 6. Syntactic Pattern Recognition: An Introduction, R.C.Gonzalez and M.G.Thomason
  • 7. Pattern Classification and Scene Analysis, R.O. Duda and P.E. Hart, Wiley 1973
  • 8. Pattern Recognition Statistical, Structural and Neural Approaches, R.J Schalkoff, Wiley, 1992
  • 9. Artificial Intelligence: An Engineering Approach, R.J. Schalkoff, McGraw-Hill, 1990
  • 10. Pattern Recognition Engineering, M. Nadler and E.P. Smith, Wiley, 1993
  • 11. Digital Image Processing and Computer Vision, R.J. Schalkoff, Wiley, 1989
  • 12. The Image Processing Handbook, J.C. Russ, CRC Press, 1992
  • 13. Digital Image Processing, K.R. Castleman, Prentice Hall, 1996
  • 14. Digital Image Processing, R.C.Gonzalez & R.Woods, Addison-Wesley, 1992
  • 15. Digital Image Processing, W.K. Pratt, Wiley 1991
  • 16. Fundamentals of Digital Image Processing, A.K. Jain, Prentice Hall, 1989
  • 17. Digital Picture Processing, A. Rosenfeld and A.C. Kak, Academic Press 1982
  • 18. Digital Pictures, A.N. Netravali and B.G. Haskell, Plenum Press 1988
  • 19. Vision in Man and Machine, M.D. Levine, McGraw Hill 1985

Journals

  • 1. IEEE Transactions on Image Processing
  • 2. IEEE Engineering in Medicine and Biology
  • 3. Computer Vision, Graphics and Image Processing (CVGIP)
  • 4. CVGIP: Graphical Models and Image Processing
  • 5. CVGIP: Image Understanding
  • 6. IEEE Transactions on Medical Imaging
  • 7. Computerized Medical Imaging and Graphics
  • 8. IEEE Transactions on Pattern Analysis and Machine Intelligence
  • 9. IEEE Transactions on Computers
  • 10. Pattern Recognition
  • 11. IEEE Transactions on Signal Processing
  • 12. IEEE Transactions on Neural Networks
  • 13. IEEE Transactions on Geoscience and Remote Sensing
  • 14. Photogrammetric Engineering and Remote Sensing
  • 15. International Journal of Remote Sensing
  • 16. Journal of Visual Communication and Image Representation
  • 17. ACM Siggraph publications
  • 18. Numerous Conference Proceedings and other journals from:

IEEE - Institute of Electrical and Electronic Engineers

SPIE - The International Society for Optical Engineering

SMPTE - The Society of Motion Picture and Television Engineers

PRS - Pattern Recognition Society

ACM - Association for Computing Machinery