Help with forming a temperature image using black and white scale?

I need to form a temperature image by assigning a black and white (thermal greyscale image) scale to a temperature scale - so by matching max/min temperatures to 255/0 values respectively, e.g. min temp 7 = 0 and max temp 42 = 255 (or 1 if this is easier going forwards?). Just wondering if someone can point me in the right direction of how to put this into code as simply as possible please? The reason for doing so is to use the temperature image data in a 2D plot.

Answer 1

[Image Analyst]

If your matrix has values going from 0 to 255 and you want temperature values of 7 to 42 you can do this:

temperatureMatrix = rescale(grayLevelMatrix, 7, 42);

It's not just 0 and 255. It will put whatever the min is to 7 and whatever the max is to 42.

If your matrix has values going from 7 to 42 and you want gray levels of 0 to 255 you can do this:

grayLevelMatrix = rescale(temperatureMatrix, 0, 255);

If all you have is a pseudocolor image with a colorbar, you can adapt the attached demo.

[Image Analyst]

You simply do

TemperatureImage  = rescale(Image1, MinTemp, MaxTemp);

Since the color bar goes from 0-255 (I believe) then that's all you need to do. You do not need to rescale the image, which won't do anything if your colorbar already includes 0 and 255.

[Katie Evans]

Oh ok, that makes a lot more sense. Thank you so much for your help!

[Image Analyst]

I hope you know about emissivity. Maybe the emissivity of trees and grass might be similar. Not sure about the sky so the temp might not be comparable. Also you need to know the emissitivy that your camera was set to. If it was set to the wrong value, the temperatures won't be accurate.

[Image Analyst]

Looks like you're trying to build a visualization of two images taken of a scene

1. The thermal (gray scale, infrared, temperature) image
2. The visible color image which has been converted to HSV color space.

I'm not really sure from the plot which data points you want to extract and which you want to discard. You can try thresholding if you think slicing off slabs from that thing might reveal what you want. If there's other stuff in there and thresholding can't reveal it, you'd have to somehow define the data points you want to extract or discard. Maybe start a new question with the thermal (gray scale) temperature image, and the RGB image both attached and say what in the image you want to visualize in that 3-D scatterplot.

Answer 2

[DGM]

If you know what the exact relationship is between gray levels and temperatures (and you can presume that it's linear), why not just scale based on those knowns?

inrange = [0 255];
outrange = [7 42];

graypict = [10 250]; % pretend this is our very small image (does not span inrange)

% scales inrange to outrange
T1 = range(outrange)*(double(graypict)-inrange(1))/range(inrange) + outrange(1)

T1 = 1×2

8.3725 41.3137

% scales data extrema to outrange
T2 = rescale(graypict,outrange(1),outrange(2))

T2 = 1×2

7 42

This will result in the same scaling regardless of whether the image data spans the entire input range. This will be different than the results from rescale(), which will map the input extrema to [7 42], regardless of what they actually are. You need to decide whether that suits your needs. If you can guarantee that your data range always spans [0 255], then the two methods will be the same.

Answer 3

[yanqi liu]

yes，sir，may be use colormap to get image，such as

a = randi([7 42], 64, 1);
a = repmat(a, 1, 64);
map = linspace(0, 255, 64);
map = repmat(map(:), 1, 3);
figure; imagesc(a);colormap(map/255);