Radiometric Thermal Camera with SDK

[Fratink]

Hello,

I'm writing some drivers for thermal cameras that we're planning on using for research and I'm noticing a lack of information from FLIR about what needs to be recorded. I'm getting the raw image data, but it doesn't have any calibration information, or anything that would allow values on the image to be assigned a temperature. Has anyone worked with capturing raw data from a thermal camera? I'm curious about what information I have to grab from the camera in order to make sure the data is useful.

Cheers,

 

[IRstuff]

Generally, that tends to fall into proprietary information, as that's presumably the very information that provides the value-added functionality of radiometric accuracy.

In general, the camera needs a minimum of calibrations against a uniform source at two temperatures. These establish a linear regression against the output counts. The offset is simply residual noise and radiometric background. The slope is the responsivity, typically, counts/radiance. If your camera outputs radiometric temperature readings, then it's also doing a inversion of measured radiance to get a radiometric temperature. If an emissivity is selected or defaulted, then the radiometric temperature is an estimated thermometric temperature. Because no detector is perfectly linear over its dynamic range, multiple calibration temperatures are usually collected so that the responsivity can be corrected over the detector's dynamic range.

For uncooled bolometric cameras, the thermal environment significantly impacts the apparent radiance measured by each pixel, and multiple calibration and camera temperatures are used to select the correct responsivity as a function of camera temperature and measured scene radiance.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[Fratink]

Hello IRstuff,

This is a scientific/industrial camera designed to be used with custom built software compliant with genicam, so I'm expecting I can get everything I need (just not in as pretty an interface as what they sell).

I had another discussion on the EEVblog forums and talked through the registers I needed in order to get the metadata. It turns out that you can set the camera in either radiometric or thermal-linear mode. Thermal linear mode then returns an image who's individual pixels have values in K. That mode is possibly best and creates the most useful "images". It should be possible to go between radiometric and thermal-linear mode in post if you know all the coefficients used to calculate the temperatures. I think you can access the calibration info as well through the genicam interface (but don't quote me on that).

As to your comments on accuracy, yes, that's one thing we're realizing that we have to communicate to the researchers is that absolute accuracy is difficult to achieve, especially with varying conditions like those found outdoors. Scientific visual imagery is difficult and we're learning that thermal imagery is even worse since it behaves so very differently.

Cheers,

 

[IRstuff]

We're actually in the midst of developing an infrared radiometer, and the calibration is a huge deal. The only saving grace is that our operating environment will be very limited, so the calibrations required are somewhat fewer than for a general purpose radiometer.

I would also caution you to check the lens; the commercial uncooled sensor we purchased appears to have either haze or veiling glare in the optics, which can seriously mess with both spatial resolution and radiometric accuracy.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm