Author Topic: Calibrate image acquisitions?  (Read 4615 times)

Anette von der Handt

  • Global Moderator
  • Professor
  • *****
  • Posts: 298
    • UMN Probelab
Calibrate image acquisitions?
« on: September 29, 2014, 04:46:44 PM »
Hi John,

is there a good way to calibrate the images that I acquire through "digitize image". After the last service everything is a little off. I am aware that I can calibrate individual images after acquisition but I wonder about a sort of machine calibration (there is a length/width of field setting in the probewin.ini file, something there). I couldn't quite find a good recipe in the manual.

Thanks!
Anette
Against the dark, a tall white fountain played.

Probeman

  • Emeritus
  • *****
  • Posts: 2502
  • Never sleeps...
    • John Donovan
Re: Calibrate image acquisitions?
« Reply #1 on: September 29, 2014, 05:07:18 PM »
Hi Anette,
Now that the mosaic imaging is working so well:

http://probesoftware.com/smf/index.php?topic=324.msg1700#msg1700

I will be turning my attention to making the image calibration more automatic. Right now the Probewin.ini file supports X, Y and rotation calibrations based on keV and mag as described in this post:

http://probesoftware.com/smf/index.php?topic=138.msg1429#msg1429

But you have to edit the Probewin.ini values by hand. So that is what we have for now, but I will keep you updated on the new image calibration feature as it gets closer. I should probably have it finished by next week (I hope). The idea is that you load a stage image and a beam scan image that overlap significantly and identify 3 points in both images that are correlated. Then the program will save the calibration automatically based on the image kev and mag...
The only stupid question is the one not asked!

Anette von der Handt

  • Global Moderator
  • Professor
  • *****
  • Posts: 298
    • UMN Probelab
Re: Calibrate image acquisitions?
« Reply #2 on: September 29, 2014, 06:00:19 PM »
Thanks! I thought that I read somewhere a good walk-through for it but could not find it anymore and mis-trusted my memory.

I will do that then.

Against the dark, a tall white fountain played.

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 2966
  • Other duties as assigned...
    • Probe Software
Re: Calibrate image acquisitions?
« Reply #3 on: September 30, 2014, 09:26:38 AM »
Thanks! I thought that I read somewhere a good walk-through for it but could not find it anymore and mis-trusted my memory.

I will do that then.

This is from the documentation. 

ImageInterfaceImageIxIy=1.0
This parameter is used to specify the aspect ratio of X over Y, for the normal beam scan generated by the imaging interface. A value of one is the default and would specify a square beam scan. A value greater or less than one would specify a rectangular beam scan. Normally the SX100 Video interface is a 4/3 ratio (or 1.333), so for example 1024 x 768.

ImageInterfaceCalNumberOfBeamCalibrations=1
This new field specifies the number of beam scan calibrations that are utilized by the software to calibrate the beam scan images. The minimum (default) is 1, the maximum is 32.

Basically the beam scan calibration is based on the current keV and magnification and returns the X or Y scan width or height and the scan rotation. If only one beam scan calibration is specified, the application always utilizes that value adjusted for the current magnification assuming a linear 1:1 relationship in magnification.

If more than one beam scan calibration is specified, the values must be separated by commas and enclosed in double quotations as shown here:

ImageInterfaceCalNumberOfBeamCalibrations=2
ImageInterfaceCalKeV="15.0,15.0"
ImageInterfaceCalMag="400.0,800.0"
ImageInterfaceCalXMicrons="805.0,402.5"
ImageInterfaceCalYMicrons="803.0, 401.5"
ImageInterfaceCalScanRotation="0.0,0.1"

The program will search for matching keV calibrations. If none are found it will use the first beam scan calibration.  If only one is found to match the specified keV, it will perform a linear extrapolation based on a 1:1 relationship in magnification.

Otherwise, if multiple calibrations are found the program will find the two nearest neighbor beam scan calibrations for the specified keV and interpolate them.

ImageInterfaceCalKeV=”15.0”
This parameter specifies the operating voltage in kilovolts used for the ImageInterfaceCalXMicrons and ImageInterfaceYMicrons calibration. The default is 15 and the minimum is 1 and maximum is 50.

ImageInterfaceCalMag=”400.0”
This parameter specifies the magnification used for the ImageInterfaceCalXMicrons and ImageInterfaceYMicrons calibration. The default is 400 and the minimum is 10 and maximum is 10000.

ImageInterfaceCalXMicrons=”805.0”
ImageInterfaceCalYMicrons=”803.0”
These parameters specify the actual X and Y beam scan size at the kilovolt and magnification parameters specified above. These parameters are used the calculate the stage position of the beam deflection. The default ImageInterfaceCalXmicrons is 800 and the default ImageInterfaceCalYmicrons is 800 and the minimum is 10 and maximum is 10000.

ImageInterfaceCalScanRotation="0.0"
This parameter specifies the retruned or interpolated scan rotation for then specified keV and magnification. The default is 0 and the minimum is 0 and maximum is less than 360.

I never really documented this very well because first, our installers always do this for you and second, I've always intended to make this much more automatic.

Watch this space next week for the new developments...
« Last Edit: February 24, 2018, 12:31:24 PM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Probeman

  • Emeritus
  • *****
  • Posts: 2502
  • Never sleeps...
    • John Donovan
Re: Calibrate image acquisitions?
« Reply #4 on: November 17, 2016, 12:32:47 PM »
For calibrating the beam scan for image calibration to stage coordinates and also for PictureSnap, the preferred method now is to utilize a stage scan image (BSE or SE), and a number of beam scan images at different magnifications as described here:

http://probesoftware.com/smf/index.php?topic=396.0

This method utilizes three fiducial reference points to more accurately calibrate the beam scan, but also to calibrate the scan rotation.

One can still utilize the original two point method described here, but it is less accurate and doesn't allow one to calibrate the scan rotation:

http://probesoftware.com/smf/index.php?topic=138.msg1429#msg1429
« Last Edit: November 17, 2016, 03:14:08 PM by Probeman »
The only stupid question is the one not asked!

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 2966
  • Other duties as assigned...
    • Probe Software
Re: Calibrate image acquisitions?
« Reply #5 on: February 24, 2018, 09:11:19 AM »
Although it is more accurate to utilize the three point image calibration in CalcImage as described here:

http://probesoftware.com/smf/index.php?topic=396.0

We decided to improve the ease of use for the simple two point image calibration in the PFE Imaging window (accessible by clicking the tiny @ button), by modifying the code to allow the new image calibration parameters to be automatically saved from the PFE image calibration dialog as seen here:

« Last Edit: April 13, 2020, 10:19:31 PM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 2966
  • Other duties as assigned...
    • Probe Software
Re: Calibrate image acquisitions?
« Reply #6 on: February 25, 2018, 11:45:36 AM »
We decided to improve the ease of use for the simple two point image calibration in the PFE Imaging window (accessible by clicking the tiny @ button), by modifying the code to allow the new image calibration parameters to be automatically saved from the PFE image calibration dialog as seen here:

In case anyone doesn't know, the image (two point) calibration window in PFE as accessed from the Imaging window (from the Acquire! window) by clicking this tiny "@" button here:

« Last Edit: April 13, 2020, 10:09:22 PM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 2966
  • Other duties as assigned...
    • Probe Software
Re: Calibrate image acquisitions?
« Reply #7 on: February 27, 2018, 10:01:47 AM »
The latest version of Probe for EPMA (v. 12.1.9) now displays the current image calibrations in the image calibration window:



I find it useful to compare the new calibration with existing calibrations.
« Last Edit: April 12, 2020, 07:48:21 PM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Rom

  • Professor
  • ****
  • Posts: 76
Re: Calibrate image acquisitions?
« Reply #8 on: June 18, 2022, 08:43:09 PM »
Good day.
When we collect BSE image in "Digitize Sample Positions On Image" window, save the image as BMP, and open the image from PictureSnap, the image is already automatically calibrated (2-points calibration).
 Could you explain what data the system uses for automatically calibration the image:
information about central point coordinate and  magnification of the image or something else?

Thank you.
« Last Edit: June 19, 2022, 08:50:38 AM by John Donovan »

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 2966
  • Other duties as assigned...
    • Probe Software
Re: Calibrate image acquisitions?
« Reply #9 on: June 19, 2022, 08:39:14 AM »
When we collect BSE image in "Digitize Sample Positions On Image" window, save the image as BMP, and open the image from PictureSnap, the image is already automatically calibrated (2-points calibration).

Could you explain what data the system uses for automatically calibration the image:
information about central point coordinate and  magnification of the image or something else?

Great questions.

The analog signal images acquired in Probe for EPMA in the Imaging and/or Digitize Image windows are calibrated based on the current (center) stage position and the magnification calibration which is stored in the probewin.ini file in the [Image] section as seen here for a typical JEOL instrument:

ImageInterfaceCalNumberOfBeamCalibrations="4"
ImageInterfaceCalKeV="20,20,15,15"
ImageInterfaceCalMag="200,400,400,1000"
ImageInterfaceCalXMicrons="454.5161,224.7639,226.9317,89.51028"
ImageInterfaceCalYMicrons="444.3432,222.1757,222.5682,89.19152"
ImageInterfaceCalScanRotation="0,0,0,0"

Note that the scan rotations were not calibrated.

I assume you are asking about this because when I was working with Ying Yu the other day I noticed that her Probewin.ini file only had a single magnification calibration at 400x.  I mentioned to her that I suspect this was merely a nominal calibration and not calibrated specifically for her instrument.

To calibrate ones instrument, these image calibrations are performed using the Calibrate Electron Image calibration window which is accessed from the Imaging button window, which is in the Acquire! window. To perform these calibrations one should use a sample with a number of small but unique features which can be located in both the optical display and also visible in the acquired BSE or SE image. Once a suitable image has been acquired in the Imaging window, one should click on the little "@" button in the upper  left of the image and the Calibrate Electron Image window is opened as shown in the previous post by myself.

In a perfect world, only a single calibration at a single magnification would be necessary. But we don't live in a perfect world!

I don't know the scan coil ranges on the JEOL 8200/8500 (where the magnification switches to different circuits), but ideally these different ranges are calibrated by your service engineer to obtain a smooth calibration as the magnification is changed.  But since these hardware calibrations do drift out of adjustment over time, we have created this calibration adjustment capability in software. Here is an example from my SX100:

ImageInterfaceCalNumberOfBeamCalibrations="5"
ImageInterfaceCalKeV="15,15,15,15,15"
ImageInterfaceCalMag="1335,758,571,405,334"
ImageInterfaceCalXMicrons="282.546,498.7495,659.7579,926.9622,1118.275"
ImageInterfaceCalYMicrons="207.0935,376.3542,504.1712,705.2865,848.4153"
ImageInterfaceCalScanRotation="9.3,9.1,8.9,8.8,8.7"

The above example shows the scan rotation calibration changing as the magnification changes.  This is not unusual, but what is a little unusual is that our scan coil is accidentally rotated about 9 degrees out of orthogonality.  This scan rotation needs to be checked at each magnification using the edge of your sample holder to be sure you are orthogonal in X or Y.  You'll also want to set the default scan rotation for your instrument (not available in the JEOL 8900) using this value in the [hardware] section of the Probewin.ini file:

ScanRotationPresent=1      ; non-zero = scan rotation interface present (Cameca SX100 and JEOL 8200/8500 only)
ScanRotation="9"                    ; default scan rotation value to send on startup (01-17-2011)

Then as mentioned above, acquire an image with two features on a diagonal (near two corners of the image) and proceed to locate the feature by moving your stage and then selecting the corresponding pixel using the mouse.  Then clicking the Calculate Image Calibrate button followed by the Save Image Calibration to Probewin.ini if the calibration looks good.

Remember, to check your scan rotation for each image acquisition!

Generally one should calibrate two magnifications for each scan coil range.  I've asked one of our consultants to write up a procedure for this process but this post should provide you with the general idea.

When your magnification values are sufficiently calibrated one should be able to utilize the Digitize Image feature and obtain accurate stage positions from your acquired images, with the possible exception of very low magnifications which exhibit significant distortion on most instruments.

Also be sure to utilize a slow scan speed for these magnification calibrations- generally the same slow scan speed you utilize for your Digitize Image acquisitions.
« Last Edit: June 19, 2022, 05:23:50 PM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Karsten Goemann

  • Global Moderator
  • Professor
  • *****
  • Posts: 218
Re: Calibrate image acquisitions?
« Reply #10 on: August 29, 2022, 10:32:39 PM »
Hi All,

I've put together a short tutorial how to use the new, more automated image calibration procedure in the Imaging window, see attached PDF.

Cheers,
Karsten