MedianDetectorTest should allow for specific levels of the instrument

[Peter Peterson]

Since it is frequently desired to compare to the median value of the tube/bank/row rather than the full instrument, add a parameter that allows the user to select this with the default being the whole instrument. The parameter should specify the number of levels above pixel to calculate the median as an integer.

[Andrei Savici]

MedianDetectorTest allows testing by tube. Refs #4899

Changeset: 95aba4f928a4f87b4fe5fe752babbe818c245585

[Andrei Savici]

Fix Windows int vs size_t for MedianDetectorTest. Refs #4899

Changeset: d0e4cadd760541a50f99b0ab53250c4e5f97b891

[Peter Peterson]

This needs to be integrated into the DetectorDiagnostic algorithm before it can be closed. A couple of problems I have already had:

• DetectorDiagnostic already integrates the workspace (once to save everybody time) and the InputWorkspace validator does not like it.
• It doesn't appear to work for NOMAD data when LevelsUp=1. The error it produces is (with a nice typo) "no single valid histohrams identified in the workspace".

[Andrei Savici]

Median set to 0 instead of throwing. Refs #4899

Changeset: 97f3695ba561b16d1aaeefb40c4079d231ff3354

[Andrei Savici]

Median set to 0 instead of throwing. Refs #4899

Changeset: 97f3695ba561b16d1aaeefb40c4079d231ff3354

[AndreiSavici]

Lower log level for calculating median. Refs #4899

Changeset: fc886e18a135d251b1774f364e84d57d60dede38

[AndreiSavici]

Lower log level for calculating median. Refs #4899

Changeset: fc886e18a135d251b1774f364e84d57d60dede38

[Andrei Savici]

Wiki will be updated during the testing period. No unit tests for this new feature.

[Andrei Savici]

Updated wiki for MedianDetectorTest. Refs #4899

Changeset: aa3dd5ad87dbd5c4a4d79a1acc7ba84cd495f681

[Martyn Gigg]

I can't test this with any ISIS data we have as it doesn't seem to cope with grouping of detectors, it simple throws an error. Not required for ISIS at the moment so see #5241

[Janik Zikovsky]

Made ticket #5245 so you wouldn't forget to add a unit test :)

[Janik Zikovsky]

Ticket #5074 really needs to be resolved for this algo to be considered working. Also, with LevelsUp=1 the results looked funny. Will attach screenshot.

[Andrei Savici]

Ticket #5074 fixed the masking problem. The funny image is due to the fact that you need to exclude zeros from median.

[Janik Zikovsky]

Works if you exclude zeros or use more data than the small CNCS file I was using.

[Andrei Savici]

Updated wiki for MedianDetectorTest. Refs #4899

Changeset: aa3dd5ad87dbd5c4a4d79a1acc7ba84cd495f681

[Stuart Campbell]

This ticket has been transferred to github issue ​5745