GSoC

This page collects the topics for Google Summer of Code, and provides brief information about them.

If you have your own idea please contact us on nkp-gsoc@googlegroups.com.

Topic 1

Migrating the existing web application from Smarty (PHP) to GWT or Vaadin.

• required skills:

Java, Maven, Vaadin, Git, (PHP)

• description:

Firstly, the student is going to elaborate the design of the GUI, preserving all the existing functionality. Subsequently, according to design the implementation phase can begin. Student should use Java platform including RIA framework Vaadin. Part of the project is also the providing a way to invoke the existing tools written in C/C++ using JNI or Java Runtime, or, if possible, rewrite these tools to Java. At the end, the documentation should be also provided. We prefer the form of wiki-based documentation. If you are interested in this topic contact me directly and send also a link to your existing project / work.

• mentor:

Jiri Kremser (jiri.kremser@gmail.com)

Topic 2

Multithreaded implementation of new similarity metrics (image data) into existing framework.

• required skills: C / C++, Java, Git

• description: The functionality of project Differ is not only validation and identification of still image file formats, but also the ability to compare two different image files and processing of such comparison. For this purposes we would like to introduce some new ways helping the comparison, namely, SSIM (http://en.wikipedia.org/wiki/Structural_similarity), PSNR (already implemented, but is single threaded), MSSIM and "A Universal Image Quality Index". If you have mathematical background and you want to work on real problems, feel free to contact us.

• Preferred is implementation in OpenCV library (http://opencv.willowgarage.com/wiki/).

• links:

  • https://ece.uwaterloo.ca/~z70wang/publications/ssim.html
  • https://ece.uwaterloo.ca/~z70wang/publications/msssim.pdf
  • Z. Wang, A.C. Bovik, "A Universal Image Quality Index", IEEE Signal Processing Letters, vol.9, no.3, pp.81-84, Mar 2002.

• mentor:

Vaclav Rosecky (xrosecky@gmail.com)

Topic 3

Detection of inconsistent image metadata and of badly scanned image data

• required skills:

C / C++, Git

• description:

• Check the color space in the file’s header

  • Grayscale = Grayscale(1)
  • or
  • RGB = RGB(1)

• Check the actual number of occupied channels in the file’s body (bitstream of image)

  • One channel in use of RGB = OK - Grayscale(2)
  • Two channels in use of RGB = ERROR
  • Three channels in use of RGB = OK - RGB(2)

• Comparison of detected Color Space values

  • Grayscale(1) = Graycale(2) = OK
  • RGB(1) = RGB(2) = OK
  • Grayscale(1) ≠ Graycale(2) = ERROR
  • RGB(1) ≠ RGB(2) = ERROR

• GLITCH inspection

Measuring of RGB channel levels

* R-255-60-100%=FALSE (unusual color coverage)
* G-255-60-100%=FALSE (unusual color coverage)
* B-255-60-100%=FALSE (unusual color coverage)
* RGB (R-255, G-255, B-255) 100% White color (unusual color coverage)
* RGB (R-0, G-0, B-0) 60-100% Black color (unusual color coverage)

Check the attendance of color components in the files labeled as grayscale * R-1px=FALSE (1 color pixel detected) * G-1px=FALSE (1 color pixel detected) * B-1px=FALSE (1 color pixel detected)

• mentor:

Rudolf Kreibich (Rudolf.Kreibich@nkp.cz)