very nice to see progress in this project. Some of us really need that stuff and are curious how it will look like.
I absolutely agree. It is very convenient to write
rai = Regions.union(rai1, rai2). I would furthermore support if there were overloaded versions for discrete binary RAIs. You may again copy some stuff, I recently programmed: https://github.com/haesleinhuepf/imgtools_binary_ops/blob/master/src/main/java/de/mpicbg/scf/imgtools/geometry/filter/operators/BinaryOperatorUtilities.java
Now I will just think loudly in writing: From my point of view, any region could be an annotation. But not any annotation (such as points and texts) is a region. This sounds like inheritance. But when I look back to the inheritance tree of ROIs in ImageJ1, we really need to be careful here. It would be very nice to do it right this time
Maybe I understood you wrong, but I kind of disagree here. Lines and polylines are no regions, because they are infinitely thin and thus, surround/describe/represent no area. But regions must do. Maybe, it would make sense to introduce a
Polygon as a closed collection of lines and a
PolygonRegion representing the surrounded area.
From my point of view, a point collection is no region for the same reason. But: A
PixelCollection could be, as it is just a representation of a discrete binary image A
RasterizedPolygonalChain could be a region as well. I could also imagine a class representing list of cylinders along a polyline being a region. But that’s maybe something for Version 2.0 . Let’s concentrate on the basic stuff…
Is there already some kind of prototype code for the
RasterizedPolygonalChain available online? Would be interesting to see how you plan to implement this.
Thanks for taking care for the ROI stuff. Your efforts are really appreciated