Ticket #741 (closed defect: fixed)
Opened 2012-09-28T14:38:28-05:00
Last modified 2012-10-05T19:54:50-05:00
SLIMPlugin threshold control doesn't correspond to TRI2
| Reported by: | aivar | Owned by: | aivar |
|---|---|---|---|
| Priority: | major | Milestone: |
|
| Component: | slim-plugin | Severity: | serious |
| Keywords: | Cc: | ||
| Blocked By: | Blocking: |
Description (last modified by aivar)
TRI2's threshold control is available under Window/Pre-processing. It shows the image intensity log histogram with min/max values and you can specify a min/max percentage or min/max values. With SLIM Plugin the threshold control is just a photon count, default 100, that does not correspond to the 0..255 gray level values you see when you cursor around the grayscale representation.
So, TRI2 lets you specify lower/upper and SP only lower. The values specified differ so it is difficult to set SP up the same as TRI2. This means it is difficult to compare fitted images and their histograms.
Change History
comment:2 Changed 2012-09-28T18:17:57-05:00 by aivar
- Milestone changed from slimplugin1.2 to slimplugin1.1A
comment:3 Changed 2012-10-04T17:23:24-05:00 by aivar
Updated SLIMPlugin to use IJ threshold capabilities, including showing a red mask similar to that in TRI2. The grayscale image also got updated to a 16-bit image (formerly 8-bit) and has values same as TRI2. Looks wonderful, except results still don't agree:
TRI2 File gpl1.sdt, values are 0.0..642.0, if I threshold 642..642 I still get two pixels. 643..643 makes them go away. 0..642 all pixels are shown; 1..642 only 9 pixels turn red.
SLIMPlugin Same file & values, to get the two pixels set 639..642 (BTW mine are white (255,255,255), TRI2 has gray (56,56,56)). SP 0 level is the same as 1, looks a lot like TRI2 3 (or maybe 4). The equivalents of TRI2 0, 1, & 2 levels are unobtainable.
comment:4 Changed 2012-10-04T17:29:48-05:00 by aivar
The problem is that IJ 16-bit thresholding works by converting to 8-bit (adding red mask colors to the display LUT). Solution would be to write my own mask display code. A slight mitigation of this bug would be to treat 0..642 as a special case that turns off threshold mask. Might be good enough for beta release.
IJ thresholding is opposite from what SP needs anyways, pixels between min and max inclusive are masked. SP really needs to mask pixels from 0..min and max..maxValue as in TRI2 preprocessing.