Clear Filters
Clear Filters

Finding the threshold of thermal images

2 views (last 30 days)
asaf omer
asaf omer on 15 Feb 2022
Commented: yanqi liu on 16 Feb 2022
Hello.
I have some thermal images and i wanna find the optimal thresholds,is there a command that can can help with this?
So far I didnt find something usefull.
thanks

Answers (1)

Image Analyst
Image Analyst on 15 Feb 2022
There is no optimal threshold. You have to decide what temperature range you're interested in. An image just has a bunch of temperatures in it. How could it possibly know that you're interested in the range from 240 degrees C and hotter? It can't. Only you would know that.
By the way, if you need to convert a pseudocolored image into a temperature image, see my attached demo.
  3 Comments
Image Analyst
Image Analyst on 15 Feb 2022
You can find the max of the histogram and fall down a certain amount from there, like until it gets to 2% of the max:
[counts, edges] = histcounts(thermalImage);
[maxCount, indexOfMax] = max(counts);
thresholdLevel = 0;
for k = indexOfMax : length(counts)
if counts(k) < 0.02 * maxCount
thresholdLevel = edges(k);
end
end
or you could try my attached triangle threshold method, which is good at finding the "corner" of skewed histograms.
yanqi liu
yanqi liu on 16 Feb 2022
yes,sir,may be use cumsum to get the target rate
clc; clear all; close all;
thermalImage = imread('cameraman.tif');
[counts,centers] = hist(thermalImage(:), 0:255);
counts2 = cumsum(counts);
for k = 1 : length(counts2)
if counts2(k) < 0.5 * counts2(end)
else
% find 50% rate
thresholdLevel = centers(k);
break;
end
end
bw = thermalImage < thresholdLevel;
figure; imshow(bw);

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!