Re: GSoC - N-dimensional image processing project

[Jordi Gutiérrez Hermoso]

On 27 April 2013 22:28, akhil vij <address@hidden> wrote:

> Major part of my thesis work is done via m-scripting and I am quite
> comfortable with C++ and git.

We use hg, not git in Octave, but they are similar. You shouldn't have
too much difficulty learning hg.

> I have small doubts regarding this project [N-dimensional images].
>
> File format : Could you tell the major file format I would be
> working on for this project.

This part should be irrelevant as far as I'm concerned. The image
package doesn't handle the file format handling, this is done by core
Octave. By the time the images are in Octave and the image package can
work with them, they're already Octave arrays.

If you want to work on improving handling a particular format, that's
fine too, but we offload this work to GraphicsMagick. This isn't even
done in Octave.

> Base functions : What are the major/most important base functions
> you think that should be modified for such images. I think the most
> priority should be given to imread/imwrite and iminfo.

I was thinking of the image processing functions themselves such as
bwconncomp or bweuler, not the image I/O functions. This seems like a
different project from what I envisioned, but it sounds like a fine
idea too. However, as I said, I think you may have to patch
GraphicsMagick in addition to patching Octave.

- Jordi G. H.