Modify centroiding based on real data
(INSTRM-1471)
|
|
| Status: | Done |
| Project: | Instrument control development |
| Component/s: | ics_actorkeys |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Sub-task | Priority: | Normal |
| Reporter: | karr | Assignee: | karr |
| Resolution: | Done | Votes: | 0 |
| Labels: | EngRun | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Sprint: | PreEngRun05 B, PreEngRun05 C, PreEngRun05 D, PreEngRun05 E, PreEngRun05 F, preEngRun07Sep |
| Description |
|
The performance of sep (the wrapper for sextractor) produced some unexpect results during the engineering run, with very poor performance in some frames. This needs to be investigated further, and better input parameters determined. |
| Comments |
| Comment by karr [ 17/Dec/21 ] |
|
This works best when the data are cropped to the useable region and the two sides processed separately, to avoid artifacts in the background subtraction and in the boundary between the two halves of the image. I want to use the background noise image calculated during the background subtraction for variable thresholding across the image; this isn't performing as I expected from the sep documentation and needs to be investigated further. |
| Comment by karr [ 07/Jan/22 ] |
|
I checked sep for better input parameters, and modified the code to process the two sides of the image separately. This is linked to 1479, and reading the useable regions from the yaml file. |
| Comment by karr [ 13/Jan/22 ] |
|
Current values for sep: threshold of 15, minarea=10 (the default is too small), and the defaults for filtering, etc. The over scan correction, background subtraction and centroiding are done separately for the two sides of the image. Bad columns have a very simple interpolation at the moment.
|
| Comment by karr [ 13/Jan/22 ] |
|
Pushed to 1479 |
[INSTRM-1473]