http://imagej.273.s1.nabble.com/Advice-for-GUI-plugin-development-tp5001300p5001461.html
> Hi Riccardo,
>
> On Thu, 17 Jan 2013, Riccardo Cabbri wrote:
>
>> I sent you an e-mail a few months ago about your plugin "Align Image by
>> line ROI".
> Generally it is a better idea to send such mails to the mailing list. I am
> often too busy to take care of all the Fiji issues, but thankfully others
> can help, too, if you let them.
>
> Besides, if I resolve a problem but only one person knows about it, it is
> a less good use of my time than if there is a public record.
>
>> We need to compare pictures of honeybee's brood combs to study the
>> effect on brood of treatments against varroosis (a disease also called
>> "the beekeeper's plague" for its severity) and Nosemosis type C (an
>> emergent pathogen for honeybees in europe).
>>
>> A plugin for Fiji to do so already exists, but I don't want to use it
>> because it is proprietary software (I am a free software lover and an
>> enthusiast debian user from many years).
>>
>> In these weeks I wrote a macro set that almost completely replace it,
>> the missing part is some code that align the images taken in different
>> times, so I can mark some cells on the comb as ROI and then I can locate
>> them on the subsequent pictures.
>>
>> Looking for something like this I found your plugin that works perfectly
>> for my needs except for the fact that handles only grayscale images. Do
>> you think that it is possible to obtain RGB aligned images?
> In another thread [*1*], Stephan Saalfeld pointed to the Landmark
> Correspondences plugin:
>
>
http://fiji.sc/Landmark_Correspondences>
> It is not only more powerful than the Align by Line ROI plugin, but it can
> do exactly the same things, too.
>
> By selecting two points in each image (which is basically the same as
> selecting a line) you have all you need to run the Landmark
> Correspondences plugin. It will ask you for a source image (which is the
> one that will be transformed) and a template image (onto which it will be
> transformed). The transformation method does not matter since you only
> provide two points and there is only one solution. Same for alpha and mesh
> resolution.
>
> The only thing to be careful about is the transformation class: if you
> choose "Rigid", it will only translate and rotate, but not scale.
> "Similarity" will scale, too.
>
> In case that it is more convenient for you to let the user choose line
> ROIs over point ROIs, feel free to use this macro to convert all line ROIs
> in all open images to point ROIs:
>
> -- snip --
> function lineToPointROI() {
> if (selectionType() != 5)
> return; // only works on line ROIs
> getSelectionCoordinates(x, y);
> makeSelection("point", x, y);
> }
>
> for (id = 1; id<= nImages(); id++) {
> selectImage(id);
> lineToPointROI();
> }
> -- snap --
>
> Ciao,
> Johannes
>
> Footnote *1*:
http://thread.gmane.org/gmane.comp.java.imagej/27052