IR images index scale not linear and disjunct help?!

[terin]

Hi- I used a NEC IR camera to take some images I would like analyze for mean, min, and max temperatures in different areas. While I had access to the camera's software (thermography studio) I saved all images as BMP with a color temperature scale in the corner. I thought that this would enable me to analyze the images using imagej. Yet I am having trouble with the relationship between the index value and the temperature scale. When I make a profile plot of the scale bar the plot is almost linear except at very high temperatures when the index values drop to low values again (attached?). I can't seem to get rid of this disjunct pattern no matter what I try and some of the areas within my images I need to analyze fall within the non-linear portion of this relationship. Any advice or help would be much appreciated. thank you profile+plot

[Gabriel Landini]

On Saturday 28 November 2009, you wrote:
> to analyze the images using imagej.  Yet I am having trouble with the
> relationship between the index value and the temperature scale.  When I
>  make a profile plot of the scale bar the plot  is almost linear except at
>  very high temperatures when the index values drop to low values again
> (attached?).

Without seeing the image it is difficult to know what is going on, but I guess
that the intensity (a greyscale value generated by the profile plot) of a
colour (3 values) scale bar is not the right way to recreate the calibration.

You might have to create a new image where each rbg triplet value in the scale
bar corresponds to the index value of that triplet in the LUT.
Only then you could apply a calibration if you know what the indexes
correspond to.

G.

[terin]

Thank you for the reply, much appreciated.  I read about that so I did make a custom lookup table. Unless I did something wrong. This is what I did: I exported all the RGB values in order of the scale bar and then made a custom look up table in order of the scale bar.  It helped some, making the scale more linear but then right at the higher temperatures the scale bar is still disjunct even with the lookup table.  I will try to attach an image, maybe that would help.

Gabriel Landini wrote

On Saturday 28 November 2009, you wrote:
> to analyze the images using imagej.  Yet I am having trouble with the
> relationship between the index value and the temperature scale.  When I
>  make a profile plot of the scale bar the plot  is almost linear except at
>  very high temperatures when the index values drop to low values again
> (attached?).

Without seeing the image it is difficult to know what is going on, but I guess
that the intensity (a greyscale value generated by the profile plot) of a
colour (3 values) scale bar is not the right way to recreate the calibration.

You might have to create a new image where each rbg triplet value in the scale
bar corresponds to the index value of that triplet in the LUT.
Only then you could apply a calibration if you know what the indexes
correspond to.

G.

[terin]

Here I have tried to attach an image.7210114I+copy.gif

terin wrote

Hi-

I used a NEC IR camera to take some images I would like analyze for mean, min, and max temperatures in different areas.  While I had access to the camera's software (thermography studio) I saved all images as BMP with a color temperature scale in the corner.  I thought that this would enable me to analyze the images using imagej.  Yet I am having trouble with the relationship between the index value and the temperature scale.  When I make a profile plot of the scale bar the plot  is almost linear except at very high temperatures when the index values drop to low values again (attached?). I can't seem to get rid of this disjunct pattern no matter what I try and some of the areas within my images I need to analyze fall within the non-linear portion of this relationship.  Any advice or help would be much appreciated.

thank you

profile+plot

7210114I+copy.gif

[Gabriel Landini]

On Saturday 28 November 2009, you wrote:
> Thank you for the reply, much appreciated.  I read about that so I did make
>  a custom lookup table. Unless I did something wrong. This is what I did: I
>  exported all the RGB values in order of the scale bar and then made a
>  custom look up table in order of the scale bar.  It helped some, making
>  the scale more linear but then right at the higher temperatures the scale
>  bar is still disjunct even with the lookup table.  I will try to attach an
>  image, maybe that would help.

It is not clear to me what sort of image you are dealing with, Here I assume
that the original is RGB.

My suggestion was to *recode* the pixel values to the indexes of the scale
bar.

Create a greyscale image of the same dimensions.
Now for all the entries in the scale bar of the original (asuming that the
colours are not repeated):
If red (255, 0 , 0) is the *first* colour, set all pixels with those values in
the RGB image to colour 0 (index=0) in the greyscale image.
Next colour is orange (let's say 255, 128, 0), put all pixels with that colour
to value 1 in the greyscale image and so on for the rest of the LUT entries
(256 in total).

Now you have an image which has intensity in the same order of the original
colour lut entries. You can now to apply a calibration to it using the
intensity of the scale bar. The new scale bar in your image should have an
intensity profile which is a ramp with no discontinuities from 0 to 255.

G.

[terin]

This forum was very useful.  I got the help I needed. A plugin helped me solve this issue.

terin wrote

Hi-

I used a NEC IR camera to take some images I would like analyze for mean, min, and max temperatures in different areas.  While I had access to the camera's software (thermography studio) I saved all images as BMP with a color temperature scale in the corner.  I thought that this would enable me to analyze the images using imagej.  Yet I am having trouble with the relationship between the index value and the temperature scale.  When I make a profile plot of the scale bar the plot  is almost linear except at very high temperatures when the index values drop to low values again (attached?). I can't seem to get rid of this disjunct pattern no matter what I try and some of the areas within my images I need to analyze fall within the non-linear portion of this relationship.  Any advice or help would be much appreciated.

thank you

profile+plot