[ome-devel] [ome-nitpick] ROI types
Roger Leigh
r.leigh at dundee.ac.uk
Sat Apr 28 10:52:12 BST 2012
On 25/04/2012 12:12, Roger Leigh wrote:
> On 24/04/2012 19:14, Roger Leigh wrote:
>> I've been looking at the ROI types used by the Zeiss AvioVision ZVI and
>> TIFF/XML formats, basically learning what types are available and how
>> they work, so I can work out how to represent them in the current OME
>> ROI model. I've attached a summary of the different available shape
>> types. They are split into two major categories, "annotations" (which
>> are just overlaid shapes) and "measurements" (which are exactly the
>> same, but contain additional feature lists of measurements to overlay
>> onto the images, and to store in results tables). Some more specialised
>> types such as LUT/gradient shapes are for annotation only, not making
>> sense for measurement.
>
> To follow up on this, this details some "nice to have" features which
> would let the Zeiss types be represented in the model visually (but not
> as editable objects in their own right):
One additional detail which was in the attachment, but not mentioned, is
that the Zeiss format specifies a bounding box for each shape. This can
be useful for rendering, since it gives the maximum size of the shape.
But it's also useful as an optimisation for UI/display since you can use
it to quickly determine if a given pixel is in a ROI (using e.g. a
quadtree). You only need to render the shapes whose bbox falls within
the given pixel to determine if you've clicked on it. The Zeiss UI also
uses the bbox to select multiple shapes--a shape is only considered
selected if you drag the selection rectangle around the whole bbox.
If using SVG-style or PostScript-style rendering, it's also usable as a
clip region while drawing. But obviously only useful if you want to
mask out drawing outside the box, which probably isn't the case at present.
It's of course possible to calculate the bbox when drawing each shape;
the shapes themselves contain all the necessary information. However,
this does require drawing them, and if you have a very large number of
ROIs, this may be prohibitive.
Not sure if this is something which would or would not be useful for the
OME ROI model, but thought I would just mention it for consideration.
Roger
The University of Dundee is a registered Scottish Charity, No: SC015096
More information about the ome-devel
mailing list