How to Enhance FLIR A315 Image similar to what I can do in GIMP?

[NickEng]

I have a problem detecting faulty hot mastic injections using VBAI and a FLIR A315 camera.  While I can detect large sized injection faults, small faults are not detectable, largely due to my inability to enhance the image in a fashion that would allow me to detect the smaller faults.  I have tried using the tools provided by Vision Assistant, but have not been successful, partly because of my lack of experience with the software and partly because Vision Assistant freezes up on me on a fairly regular basis and slows down my progress (which is quite frustrating and annoying).

I have saved some logged images that the FLIR A315 camera aquires (using the VBAI program I have created) and have attached them to this forum post so you can see what I am working with.

I have only been able to post 3 of 4 images as attachments to this post, so please see the continuation of this in the reply that I will add next...

[NickEng]

As you can see, the information is there in the images captured by the IR camera and when appropriately processed, it should be very easy to discern good from bad parts.

Unfortunately, when I use VBAI, I cannot achieve the same degree of enhancement as I can when using GIMP, which is photo-shop like software that has many image processing filters.

I have attached a screen capture of the GIMP filter that I used to enhance the logged VBAI FLIR camera image acquisitions.

Is there a way of achieving the same type and degree of enhancement using VBAI filters?

FWIW, the images I am getting from the FLIR camera appear as 16 bit grey scale images in VBAI.

I'd appreciate some help from a VBAI guru!

Thanks in advance,

Nick

P.S.  I can post my VBAI 2010 project file if that would be useful or helpful.

[toto26]

Hi,

As you have a 16bit image, either you choose function that works on 16bit images, either you first map to an 8 bit image.

You can easily do this with a vision assistant step, and then add a conversion function. Select "adjust dynamic".

But you might have different results depending on the min and max value of your image.

What woulb be the best is to know your extremes temperatures and map from 16 to 8 with these values.

Regards

Here is my result just using a vision assistant step

[NickEng]

I tried mapping to 8 bit with "adjust dynamic" and I got very inconsistent results as there are often cold particles or water droplets on the extrusion.

The GIMP filter I show give me the best results, but I can only apply it off-line.  It would be great if I could implement the equivalent filtering in real-time.

If you have any suggestions, they would be much appreciated.

Nick

--

[toto26]

Hi

The solution is then to call a run LabVIEW vi with an IMAQ UserLookup function inside.

Give a 65535 sized array containing the same LUT as you show in your gimp screenshot.

Hope this helps

[NickEng]

Newbie Questions: 

1. Is is possible to create a LabView vi in NI's Vision Builder for Automated Inspection (VBAI) to implement a look-up function that implements a filter to do what I show I've done with the GIMP software filter?  Or do I need to have the LabView software purchased and installed as well in order to create vi's?
2.  What about using C/C++ or C# to implement the desired vi filter?  Is another software purchase required to it that way?

Thanks!

Nick

[toto26 · Accepted solution]

You will need LabVIEW to create vi you can then call with the runLV vi step.

You won't be able to call dll or C/C++ code from within VBAI.

Regards

[Epsilon]

Hi!

We have done several applications using FLIR A315 with VBAI. The image has a 16-bit depth, but the camera is only using 14bits out of the 16. You can dynamically scale the image by detecting the hottest and coldest pixel in the image or region of interest and then transform this to an 8-bit format. Make sure your scaling applies to the data you are processing and not just to the displayed data on the screen. There are also some firmware bugs in some camera models, depending on when they have been purchased. These can lead to strange behaviors in the image scaling.

Best,

Markus