[GRASS-dev] Report #3, GSoC, GRASS Image Segmentation

Hi!

Sorry for the day late report. I was pushing to have the basic algorithm going, and wanted to report that it already works… alas, it compiles, but there is still some other bug or logic error I still need to find…

Completed:

  • I/O works, the module can read imagery groups into segmented (tiled) files. (Thanks Markus for spotting the incorrect function names!)
  • Updated pseudocode, adding details as my understanding of the processing steps (and of C) are getting better
  • Started coding the segmentation algorithm

Plan for next week:

  • Get the basic segmentation algorithm running
  • I will have a number of options and more speed/memory improvements in the code to work

Blocking/Difficulty:

  • Not stopping work, but the toughest part is considering what data structures to use, trying to balance speed for small/medium maps but also be able to have enough RAM for very large maps.

-Eric

On Sun, Jun 10, 2012 at 6:52 AM, Eric Momsen <eric.momsen@gmail.com> wrote:
...

Blocking/Difficulty:

* Not stopping work, but the toughest part is considering what data
structures to use, trying to balance speed for small/medium maps but also be
able to have enough RAM for very large maps.

Please collect this important information: perhaps as an addition to a related
Wiki page. Future programmers will be be happy to use it.

thanks
Markus