John

You are operating in a far higher orbit than me. 

I use Excel as a canvas to layout picture format thermal images.  It appeals to me for ease of picture manipulation and placement.  I suppose a good image software would work equally well.

 

Have you actually managed to place the manufacturer data file ie .sit into a workable format for excel?

Mikron Mikrospec will export imagery to excel with spreadsheets for image, pixel value and inspection areas pixel value.

---

Jan Arthur VanZelfden
Reliability Engineering
SKF Reliability
at SAPPI Fine Paper
Muskegon MI

 

 

Attached is the jpg version of a 1578 x 289 Radiometric Panorama.  The panorama is composed of 30 Fluke FlexCam 320 x 240 images.

 

I will be glad to send anyone that asks the real 32 bit tif image and a link to free software that will read it and allow line graphs etc.

 

I very much agree with Greg Stockton that stitching IR is a LOT harder than stitching visual images.  Thermal images don’t have the sharp details that are common in visual.  The North side of this building is thermally bland and was really hard to put into a panorama.  Persistence and trying a lot of different software finally produced a decent image.

 

Then I had to deal with an interior shot I really wanted to put into a thermal panorama.  It just wouldn’t go together.  So we put a team together to develop a better way to do IR panoramas.  We have a technique that works very well and is very easy.  We have been told it is patentable.  We hope to have it available soon.

 

The 26 KW Ovonics solar cell is just barely visible in the ironbow image.  The building gets 25% of its energy from that panel.

 

--
John C. Lafeber, P.E.
NCII - North Coast Industrial Imaging
4165 Stein Road
Ann Arbor, MI 48105

734-904-0870
LeanIR@gmail.com

John, I'd certainly be interested in a copy of that file and the link. I'm at opshouse@sympatico.ca.

Thanks

-John

Jan, to your question on exporting manufacturer's data, yes, I am exporting data to excel via download from the imager to the PC, then export from the manufacturer's software (in my case, Fluke Smartview) in one its allowable formats.  There are a number of choices, but TXT and CSV are the easiest for me to work with.  From there I have to control the import via separate macro to line up the data accurately in a plottable format.

My motivation for doing this wasn't originally to support merged, live radiometric data, but to permit tighter control of emissivity and reflectivity corrections on mottled surfaces.  The manufacturer's software didn't permit a background temp colder than -22, and some of my shots had the night sky reflected, which averages -40 and colder on a clear night.  I also wanted control over patches of repair material with wide variation in emissivity.    My excel application was set up to permit pixel level correction for emissivity. One of the spreadsheets shows the entire image, with each cell representing a pixel, displaying the temperature and with a colour matching the palette.  The spreadsheet itself has a controllable pallette, so far I use high contrast, ironbow and bluered for this.  From there, I control emissivity correction and apparent temp recalculation by cell reference I have to admit, though, I haven't gotten around to putting a good user interface on that yet.

Once I export the data from Excel to my plotting package, palette control is infinite,  I have roughly a dozen that I use, incorporating the standard set as well as a couple of specialised ones for work where I have to see an entire scene but want to highlight gradation in a specific segment.  Each pixel in the plot can be referenced by cross hair cursor and the temperature shows up.  The best feature of this is that I can deliver the data to my client with a plotting front end, so they can examine the data themselves without the need for expensive software.  I also allows me to retain some control of the source data in native form.  The merging option occurred within minutes of the first plot.  A rough merge up to 7 images wide takes about 10 minutes.  After that, tweaking to ensure the perspective is correct and correcting for overlaps and undershoots can take more time, depending on how well I took the originals.  A tripod helps - another little mod in my case, as the Ti25 has no tripod mount and I didn't want to do anything to the camera itself.

 I'm not familiar with the Mikron package.  Are you able to see what formats its PC software will export to?  If you can let me know, I can probably tell if it's manageable downrange.

Cheers

-John

John H. Where do you get these Plotting packages at. John I would like the link and file as well my email is jkays@snellinspections.com

---

Jeremy Kays Reliability Technician
Level II IR, MCA Level I, In Compliance with the ASNT guidelines
The Snell Group
jkays@thesnellgroup.com
http://www.thesnellgroup.com

Hi Jeremy.  I feel like I've already gotten carried away on this thread. It's fun stuff to be sure, but probably getting off topic.  I'll send you an email to cover specifics.

 Cheers

-John

John

If I follow your post correctly, take the pixel value data sheet exported from Mikrospec, assign a pallette to the range of values, and an image is constructed.  Something like conditional formatting on a much larger scale?  The macro does this for you? 

Is this something that is proprietary to your inspection product or are you at liberty to share?  I am quite interested from the aspect of overlap and image fit.  It seems by controlling cell size one might be able to counter some distortion issues or on the flip side become very creative!

If you are willing to share, fire away to vanzej@yahoo.com.

Thanks for the great explanation!

Jan

---

Jan Arthur VanZelfden
Reliability Engineering
SKF Reliability
at SAPPI Fine Paper
Muskegon MI

Hi Jan.  Here are three images, in order of processing (I hope they survive).

The first is a direct PNG from the manufacturer's software.

The second is how the image appears mid-point, in the spreadsheet.  Cell colour is not conditional but is handled by macro to afford controllable colours of choosable palettes. The third is the image as rendered in the final stage plot.  This process, end to end, takes about 5 minutes (4 of them waiting for the macro to run, it is resource intensive on a large image).

To your comment about cell expansion and contraction for overlap.  As you can see, the Excel version is very coarse,  This is due to the fact that the cells are rectangular, while most plots use a different shape for pixels.  This creates the equivalent of a highly aliased image in Excel. I don't use this image for anything other than locating overlap joints, the thinking being that I can drag and drop image data directly into Excel.  As it is, I have this functionality in the plotting package anyway.  Where it gets to be important is where I have to locally adjust emissivity.

As for sharing the code, at this point I am reluctant as it is entirely a personal prototype and I can't vouch for unintended effects.  I have been discussing this stuff with Fluke in the hopes that we can agree on a supportable add in.  We'll see where that goes. 

Jan, you may be interested in John Lafeber's files as well.  I've just gone through his email and his solution might be less macro intense.  Are you able to share what file formats Mikron's software will export?  A simple reformatting routine for Excel is sharable.

My apologies to the moderators for all the bandwidth I'm sucking up here.

-John

Looks like only one file per post at this size.  The first was the raw image.  Here is the excel spreadsheet

 

Final plot.  I didn't quite match the temp range in my haste, sorry.

 

John Horrigan:

Looks like only one file per post at this size.  The first was the raw image.  Here is the excel spreadsheet

 

John

This looks exactly like a surface chart rotated to a 2D view.

---

Jan Arthur VanZelfden
Reliability Engineering
SKF Reliability
at SAPPI Fine Paper
Muskegon MI

Jan, that is my intent.  I forgot to mention that, on the PDF, the black spots everywhere are the temps.  Of course, they don't show up too well on the attachment here. 

Most plotting packages will plot a suface XYZ chart.  In my case, Z is temp.  Here is another view of the same plot, this time with a different azimuth and lighting angle.  This is sometimes useful for looking at an area of interest.  You can also scale Z (temp) to highlight things even more.

The tough part is actually deciding when it's fun and when it's work...

-John