1 / 25

Zooming in on A Transformed Image: an Undergraduate Project

Zooming in on A Transformed Image: an Undergraduate Project. Caroline Haddad, SUNY Geneseo haddad@geneseo.edu Dawit Haile, Virginia State University dhaile@vsu.edu Helmut Knaust, University of Texas at El Paso hknaust@utep.edu. Outline. Background Wavelets Workshop 2006

ban
Télécharger la présentation

Zooming in on A Transformed Image: an Undergraduate Project

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Zooming in on A Transformed Image: an Undergraduate Project • Caroline Haddad, SUNY Geneseo • haddad@geneseo.edu • Dawit Haile, Virginia State University • dhaile@vsu.edu • Helmut Knaust, University of Texas at El Paso • hknaust@utep.edu

  2. Outline • Background • Wavelets Workshop 2006 • Module Writing Workshop 2008 • Project • How to zoom in on region of interest • How does the transform change the image? • Solution • Mathematica demonstration

  3. Collaborative Research: A Phase II Expansion of the Development of a Multidisciplinary Course on Wavelets and Applications (DUE-0442684)September 2007 – June 2010Patrick Van Fleet, Project Director, University of St. Thomaspjvanfleet@stthomas.eduCatherine Beneteau, co-PI, University of South Floridacbenetea@cas.usf.eduCaroline Haddad, co-PI, SUNY Geneseohaddad@geneseo.eduDavid Ruch, co-PI, Metropolitan State College of Denverruch@mscd.edu

  4. Project Development Workshop • Summer 2008, Madison, Wisconsin • Our group: • Caroline Haddad, SUNY Geneseo • Dawit Haile, Virginia State University • Helmut Knaust, University of Texas at El Paso

  5. The Project: Zooming in on a Region of Interest Problem statement: Given an image transformed by the k iterations of a Wavelet Transform, write an algorithm that finds the inverse of a sub-matrix in the blur portion without inverting the entire matrix. This is known in the medical field as a "region of interest”.

  6. Prerequisites • Basic Matrix Algebra, Haar Wavelet Transform as Applied to a Matrix or Vector, and its Inverse • Possibly another Wavelet Transform (such as one of Daubechies)

  7. Possible Areas of Introduction • Linear Algebra • Wavelet Course • Modeling Course (with a background in linear algebra) • Imaging Course • Signal Processing Course

  8. Learning Objectives or Outcomes 1. Students will gain better understanding of the inner workings of the DHWT.

  9. Learning Objectives or Outcomes 1. Students will gain better understanding of the inner workings of the DHWT. 2. Students will improve their programming skills

  10. Learning Objectives or Outcomes 1. Students will gain better understanding of the inner workings of the DHWT . 2. Students will improve their programming skills 3. Students will learn practical applications such as "region of interest" in medical imaging, e.g. in MRI's and CT scans, and “un-transforming” small portions of large audio files

  11. Instructor Deliverables • Project Description, and Mathematica or Matlab Codes • Student Deliverables • Students will submit a module to do this and a report explaining the solution approach, and why any anomalies occur. • Possible Future Work • Future work: Try to get rid of edge effects for D4, D6. Can it be generalized for even D filters?

  12. One Possible Project Outline 1. Define and identify where in the transformed matrix to obtain "region of interest". 2. Identify any restrictions on the indices of the original matrix and indices of the sub-matrix (region of interest), if any. • Given a matrix transformed by one iteration (k = 1) of the HWT, decide what other portions of the transformed matrix will be needed to invert “the region of interest”.

  13. One Possible Project Outline 4. Write a code to obtain the inverse. 5. Test it //This is done if the original picture document is known.

  14. Depending on the Student and Time… • Repeat steps 2 - 5 for k = 2, and 3. 7. Repeat steps 2 - 5 for k arbitrary.

  15. Depending on the Student and Time… • How about using D4? D6? • What do you expect to happen? What does happen when you attempt to invert the region of interest of the given transformed image? • Why do you think this happens? 11. With D6 you get weird edge effects on the top and left. For k = 1, 2, 3, the number of rows/columns affected for each is 4, 12, 28, respectively. How can you "fix" this?

  16. Discrete Haar Wavelet Transform Matrix

  17. Applying WN to an Image

  18. Applying DHWT to Image

  19. Block Form of the Transform Matrix

  20. Region of Interest

  21. Region of Interest

  22. Region of Interest  Zoom In

  23. Image after 2 iterations of DHWT

  24. Mathematica Code and Results http://baire.utep.edu/Zoom/

  25. Any questions? • Contact Information: • Caroline Haddad, SUNY Geneseo • haddad@geneseo.edu • Dawit Haile, Virginia State University • dhaile@vsu.edu • Helmut Knaust, University of Texas at El Paso • hknaust@utep.edu Thanks to Pat Van Fleet. Thanks to . Thanks to

More Related