How to convert thermal image(From FLIR T400) into corresponding temperature in matlab?

I have a thermal image with color code range.I am new to work with thermal image.I don't know how to convert the pixel value into corresponding temperature. Please send the matlab code or give some idea about it. Thanks

[jonas]

Can you upload the image?

[Suganya K]

Sure

Answer 1

[Guillaume]

The image you get from an infrared camera should be greyscale. Typically, the camera software may use false colours to make it easier to distinguish certain temperatures but the raw image is greyscale. If you have an image with false colours already applied it's going to be more difficult to process.

Of course, without a calibration that tells you how to translate pixel intensity to temperature it is completely impossible to do the conversion. So have you got such calibration?

[Suganya K]

Thank you so much for your valuable comment. Is there is any way to find temperature variation over image? Could you please tell me about calibration?

[Guillaume]

Well, the calibration should simply tell you that intensity X is xx degrees and intensity Y is yy degrees, and you rescale the image intensity accordingly. In theory, it should be just two values (slope and offset). So, if you have the raw greyscale image and the calibration coefficients, all you have to do to get temperature is:

temperature = polyval([slope, offset], greyimage);

The calibration should be given to you by the camera manufacturer or the manual shoul include the procedure to obtain it. Alternatively, just record images of objects at known temperature (and emissivity).

If your camera doesn't give you the grey image but only a false colour image, then you'll have to follow Image Analyst demo, which basically reverses the mess created by the false colouring to get back the original grey image. In his demo, the calibration is obtained from the scale of the colour bar which is probably the least scientific way to get a calibration.

As a side note, also note that the rainbow-type colour scale used in your image and in Image Analyst is problaby the worst kind of colour map to use (see also Rainbow Color Map (Still) Considered Harmful). At least, if you used a colour map whose luminosity is proportional to the signal the conversion back to greyscale would be more straightforward (just a ratio).

[melvin infant]

Well, how can I get the values for slope and offset for calibration

[Image Analyst]

Make a plot with two points, a low temperature point and a high temperature point. Put gray level along the x axis, and actual temperature in degrees along the y axis. I think you'll know what to do. If not, then:

coefficients = polyfit(grayLevels, temperatures, 1)
slope = coefficients(1)
offset = coefficients(2)

where grayLevels and temperatures are 2 element vectors (or more if you want).

To get a predicted temperature from coefficients and a sample gray level, do

predictedTemperature = polyval(coefficients, grayLevel)

Answer 2

[Image Analyst]

See attached demo. Adapt as needed.