Ticket #498 (closed enhancement: wontfix)
Opened 2011-05-04T11:35:24-05:00
Last modified 2011-06-07T10:20:49-05:00
Speed up RGBImageTranslator?
| Reported by: | bdezonia | Owned by: | bdezonia |
|---|---|---|---|
| Priority: | major | Milestone: |
|
| Component: | Legacy Compatibility | Version: | |
| Severity: | serious | Keywords: | |
| Cc: | Blocked By: | ||
| Blocking: |
Description
RGBImageTranslator is very general in making ImagePluses from Datasets. It makes no assumptions about which axes are present (requiring only x, y, and c==3) nor in what order they appear.
Since we pretty much assume the x axis is axis 0 and the y axis is axis 1 in many places in the code this generality may not be necessary. However it does allow API consumers to send this translator complex data.
None the less the current method might be slower than we want. We can make it faster via fixing the x and y axes and iterating over the others by setting plane position via dimensional axis indices and calling getPlane() using the plane position as needed. However for nonplanar data structures this could be problematic. The current implementation is immune to such problems.
Change History
comment:1 Changed 2011-05-04T11:36:19-05:00 by bdezonia
- Summary changed from Spped up RGBImageTranslator? to Speed up RGBImageTranslator?
comment:2 Changed 2011-05-27T09:20:54-05:00 by bdezonia
- Status changed from new to closed
- Resolution set to wontfix
After further thought I think it does not make sense to change this. At best we could add a speical case for planar access and iterate grabbing planes and then doing direct pixel lookup in the returned arrays. This removes generality and adds little performance improvement.
comment:3 Changed 2011-06-07T10:20:49-05:00 by bdezonia
- Milestone changed from imagej-2.0-beta1 to biweekly-2011: Jun-06 to Jun-17