Ticket #171 (accepted defect)

Opened 10 years ago

Last modified 9 years ago

decreasing radius before stitching is not sufficient for success

Reported by: pschwarb Owned by: emmenlau
Priority: critical Milestone: XuvTools-1.8.1-final
Component: xuvtools Version: 1.8.0-beta3
Keywords: radius, small, downscale, decrease, moving, window Cc: aaron.ponti@…, patrick.schwarb@…, developers@…

Description

Testfiles on the new LSM700 Confocal systems were created to test the tiling functions but it did not work although it quit obvious for the user. I tested it with differnt advanced settings.

Attachments

lsm700 Download (19.5 KB) - added by faim 10 years ago.
lsm710test.png Download (37.2 KB) - added by emmenlau 9 years ago.
Manual Stitching

Change History

Changed 10 years ago by faim

Changed 10 years ago by emmenlau

  • cc aaron.ponti@…, patrick.schwarb@… added

I have added you and aaron to the CC: list of the bugreport. If you do not add yourself to the CC: list, you will not be informed about updates :-)

Changed 9 years ago by emmenlau

Manual Stitching

Changed 9 years ago by emmenlau

  • milestone set to XuvTools-1.8.0-final

Dataset available at

xuvtools.org:/data/xuvtools_img/BugListData/Bug0171/

Added Bug to Milestone 1.8.0 final.

Changed 9 years ago by emmenlau

  • status changed from new to accepted

From Reporter:

Es sind einfach von Hand aufgenommene Tiles (nichts mit automatischer Aufnahme oder Koordinaten !!) Dann wurde versucht einfach die vier Tiles mit vollstitch zu stitchen. Für das Auge offensichtlich aber für die Software scheinbar nicht :-( .

Changed 9 years ago by emmenlau

  • keywords radius, small, downscale, decrease, moving, window added; Zeiss LSM removed
  • priority changed from minor to critical
  • summary changed from Obvious tiles from new LSM700 not stitched to decreasing radius before stitching is not sufficient for success
  • milestone changed from XuvTools-1.8.0-final to XuvTools-1.8.1-final

Renamed issue to reflect current changes.

Here an update on the issue: When starting with default settings, XuvTools decreases the moving_window_radius to the minimum occuring radius. For stack sizes of:

stack0001_size_um (11,320.087,320.087) = stackSize (11,512,512) * element_size_um (1,0.62517,0.62517)
stack0002_size_um (7,320.087,320.087) = stackSize (7,512,512) * element_size_um (1,0.62517,0.62517)
stack0003_size_um (8,320.087,320.087) = stackSize (8,512,512) * element_size_um (1,0.62517,0.62517)
stack0004_size_um (10,320.087,320.087) = stackSize (10,512,512) * element_size_um (1,0.62517,0.62517)

the correctly downsized radii are:

movingWindowRadius has been decreased from (8,16,16) to the minimum occurring radius (4,16,16)
fixedWindowRadius has been decreased from (16,32,32) to the maximum occurring radius (6,32,32)

This is however a problem if the stacks now have a Z-displacement:

maxTest 0 between tiles tile0001 and tile0002 is: (0,-1.77826,-237.398) um)
The tiles overlap in um from (-3.5,-158.265,77.3544) to (3.5,160.044,160.044) and in movingArrPixels from (2,3,380) to (8,511,511), that is [ 6 508 131 ] pixels.
The tiles tile0001 and tile0002 don't overlap enough.

Therefore an ideal solution is if XuvTools would be able to adjust both radii on the fly, not prior to starting the stitching.

Note: See TracTickets for help on using tickets.