How do you get temperature readings from the Lepton thermal images??

[Nicky Borg]

How do you get temperature readings from the Lepton thermal images?? In degrees Celcius?

[tz]

That is the reason the raw temperature data is 14 bit.  I'm not sure how to calibrate or find the curve, but if you could have a reference temperature (e.g. a metal surface at a controlled temperature, like a slowly heating pot of water (large thermal mass) with a temperature probe) you can read the values of the image at that known temperature and create a translation table or curve.  (I'm not sure that it is NOT linear so that 0 would represent a fixed cold temperature, 16383 a fixed high one and it would be linear between).

I don't think they are calibrated, but there should be a certain narrow range for the imagers, so if a table was created on a sample of devices, it could be "close".

[Nicky Borg]

Calibration is tricky.

I have a demo FLIR Lepton camera supplied by FLIR. With it came a square piece of black plastic. I assume it is a black-body.

In the instructions it says that you should put the camera in front of this black-body and on the software press calibrate.

Even in the Lepton Demo Camera by Pure Engineering (http://www.pureengineering.com/projects/leptondemo) there is a calibrate button.

However, unsure what next steps to take to get real temperature measurements after calibration.

[Pure Engineering]

In the case of the flir one from what I have seen in the teardowns  there is a temperature sensor measuring the shutter, guessing a thermopile. then with the shutter closes it can read the array and form a background image of sorts. This calibration image can be used for temperature calibration as well a potentially noise reduction. 

[Nicky Borg]

Would be great to have an API that does all this stuff and another API that outputs calibrated temperature measurements, for example in Kelvin so that user can apply a conversion (degrees C or F).

After all, the ultimate application of thermal cameras is the non-contact measurement of temperature.

[Levi Junkert]

Has anyone solved the calibration issue? Is the calibration data stored directly to the Lepton or should it be store in the application control device (i.e. Beagle/Raspberry Pi). I'm worried about losing calibration during a restart of the Lepton (power cycle).

[Hamlet Tu]

You need Lepton 2.5, not Lepton 2.0

You can convert RAW14 to degrees Celcius.

Ther are PT1 with Lepton 2.5, and Linux/Max Application.

https://groupgets.com/manufacturers/flir/products/radiometric-lepton-2-5

[Munkhtamir Oyumaa]

I can getting temperature from Lepton 2.0 . That 14 bit raw value convert by formula.

[Levi Junkert]

I am using the GroupGets break out board, not the USB PureThermal 1. I also have a Lepton 2.5 inserted in the board.

@Munkhtamir - Which formula are you using to calculate K, F, and C?

[Kurt Kiefer]

Simply put, with Lepton 2.5 the 14-bit value you get out is the temperature value in Kelvin * 100. 

I was going to refer you to the 2.5 datasheet (section 8.4) on the GroupGets product page (https://groupgets.com/manufacturers/flir/products/radiometric-lepton-2-5), but it looks like the wrong document was included there by accident! I've let Matt know, and he'll get that fixed up.

For now, here's the relevant information:

8.4 Radiometry Modes

The Lepton 2.5 release includes multiple options for radiometry modes that affect the video output signal:

• Radiometry enabled, TLinear enabled (default)

• Radiometry enabled, TLinear disabled

• Radiometry disabled

  The radiometry enabled mode affects the transfer function between incident flux (scene temperature) and pixel output. From an image-quality standpoint, both radiometry modes produce nearly identical performance (no change in NEDT), and either mode is appropriate for strict imaging applications. However, for applications in which temperature measurement is required, radiometry must be enabled to access the related calibration and software features, such as TLinear and spotmeter, which support these measurements. In radiometry enabled mode, enabling the corresponding TLinear mode changes the pixel output from representing scene flux in 14-bit digital counts to representing scene temperature values in Kelvin (multiplied by a scale factor to include decimals). For example, with TLinear mode enabled with a resolution of 0.01, a pixel value of 30000 signifies that the pixel is measuring 26.85°C (300.00K - 273.15K). The Lepton 2.5 configuration is intended as a fully radiometric release, therefore the factory defaults are defined to have both radiometry and TLinear modes enabled. 

[Levi Junkert]

Ah! Perfect this makes sense now. Thanks Kurt!

[Munkhtamir Oyumaa]

Temperature T = slope * (raw - 8192) + ambTemp

According to Max Ritter's work, the slope he calculated is 0.0217. I recorded the temperature of the difference between raw data and 8192 and the temperature measured by the MLX90614 to apply a polyfit using Matlab and I found the slope I got was 0.026. The result might vary when the environment temperature changes. 

[Nicolás Menoni]

What is the formula for taking into account emissivity?

To = Temperature of the object (target)

Ts = Temperature measured by the sensor (Lepton 2.5)

e = emissivity

1) Ts = e x To 

2) Ts = Fourth_root_of(e) x To

As I understand it is calculated using formula 1) but according to the  Stefan-Boltzmann Law I have the doubt if it should be calculated as in option 2).

As reference this short article could be useful.

I will appreciate any help. Thanks

[Shazwan Ramdan]

how can i add temperature reading in Qt GUI of FLIR Lepton thermal image in raspberry pi 3?

[Anders Bach]

I dont see how you can extract from your link that Ts = Fourth_root_of(e) x To. What about the (1-e) term?

Further, I dont think one should look at this problem in terms of absolute temperature (Kelvin) as you do in your interpretation of "Ts = e x To ". Surely, if all temperatures are at equilibrium, the equation doesn't apply.

Instead, you should look at it as a temperature difference to a reference temperature.

If you linearize the equation in your link around a Tref, such that Tx = (Tref + dTx) and discard all higher terms of dTx, you end up with dTs = e x dTo

Discarding all higher terms (2,3 and 4th order) is a good approximation when object and background temperatures are close to each other. When they are not, the system gets highly non-linear which is inherent in all calibration curves you see.

:-)

Anders

[honggang wang]

Hi Munkhtamir, 

Thanks for your equation.

One more question, if the T in your equation in K or C? Thanks.

在 2017年1月9日星期一 UTC-5下午9:08:01，Munkhtamir Oyumaa写道：

[Nati Barchilon]

בתאריך יום רביעי, 26 בנובמבר 2014 בשעה 15:28:57 UTC+2, מאת Nicky Borg:

How do you get temperature readings from the Lepton thermal images?? In degrees Celcius?

if you have lepton 3.5 or 2.5 you just devide the number by 100 and reduce it by 273.15

for example 30000 is actually 300.00 kelvin . 

so deiveide it by 100 and reduce 273.15 youll get 26.15 

 

[Hanif Izzudin]

i understand waht you mean, but i dunnt know to get a 14 bit
can you give me the source code full?
thx before...

[Choon-Ching Ng]

How about lepton 3.0? is it using the same formula as lepton 3.5 or 2.5?

[Andrew V. Jones]

The 3.0 Lepton isn’t “radiometric”, so I am not sure you can get
the “real” temperature readings, only ones that are relative the
rest of the frame.

Cheers,

Andrew

> --
> You received this message because you are subscribed to the Google
> Groups "Flir Lepton" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to flir-lepton...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/flir-lepton/dc8bb7af-9e6e-4688-a863-217c514014db%40googlegroups.com.

[Hanif Izzudin]

Hello Nati Barchilon...
I get a problem. because, when i get image from Lepton 2.5 (that have Radiometric feature, so it can be read the temperature), there is no information about a value that have you said, like "example 3000". Because, that i get is image picture that have pixel 80 x 60.