[ImageJ-devel] Results from yesterday's imglib coding session

[Curtis Rueden]

Hi everyone,

Yesterday the ImageJDev.org developers spent a few hours together,
attempting an initial integration of imglib with ImageJ.

We ended up (partially) implementing a new ImageProcessor subclass called
ImgLibProcessor that is backed by an mpicbg.imglib.image.Image<T extends
ComplexType<T>> object. This basic proof of concept is available from our
Git repository:

http://imagejdev.org/trac/imagej/changeset/3b073b3a34fae55c6e8aa0de9da210c54d558054

For those interested, the ij.process.ImgLibProcessor class has a main method
to test the functionality; the image used must be grayscale, and either a
single image or a Z-series, due to current limitations in imglib's LOCI
Bio-Formats image importer.

We also identified several next steps to take to further pursue this work:

1) Resolve the multichannel issue/limitations

2) Implement the many unimplemented imglib processor methods

3) Investigate the ImageStatistics subclasses—maybe create an analogous
ImgLibStatistics class?

4) Identify the places within ImageJ where type-specific case logic is used
  * Add a case for the new imglib backed structure at these locations

5) More generally, add getters/setters for all public fields throughout
ImageJ, to assist with delegation logic in the future

Overall, the coding session was a success—people felt like they learned
quite a bit—so we will probably do another one in 2-4 weeks, assuming we
have another appropriate topic.

-Curtis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://imagej.net/pipermail/imagej-devel/attachments/20100723/4547ce8a/attachment.html>