Color Feature Extraction of Thermal Image

I have a set of thermal images. From this set, for one image, I have coded for color feature extraction with HSV. Using HSV is best idea or should I use something else? If something else, what is the basic reason behind it?

Answer 1

[Image Analyst]

Your question doesn't have anything to do with the subject line or make too much sense. Thermal images don't have color. They are gray scale/indexed images. You can apply any colormap you want, and a lot of them give false "edges". Steve Eddins wrote a blog series on this. So basically you can use whatever colormap you think looks good but do NOT do feature extraction based on the color image. You MUST do it based on the original floating point grayscale image, which is the temperature values directly.

[Shrikanth]

My thermal image is in Lava palette. I am thinking to extract color features for determining number of combinations available in image and from it, construction of Color thermal Scaling as series of thermal images show change in color. Will this do?

Some of thermal images showing color change.

[Image Analyst]

I've added your picture to your initial post up top. It could be possible to find things based on color but it's not preferable. Don't you have the original image before the pseudocolors were applied? You should. And you should do image analysis on that image, not the color image. Anyway, I don't know what a "combination" is - can you outline it in the image or somehow indicate exactly what it is? We can't extract it from the image if we don't know what it is. Is it the purple square?

[Shrikanth]

Purple area is the ROI of which I would like to extract temperature. Above image is extracted from an encrypted file captured by thermal imager. Color of ROI will change towards Florescent gradually. This change will be proportional to color change.

[Image Analyst]

Threshold to find a mask, then get the mean

mask = thermalImage > 25;
meanTemperature = mean(thermalImage(mask));

[Image Analyst]

So whatever happened with this? Did you ever get it working? Do you still need more help?

[janifal]

[sorry for my bad english] Hi Image Analyst. First of all, an enormous compliment towards your enthusiasm in image processing. I'm a big fan of your work here.

I'm interested in understanding more about thermal images, especially in its texture analysis.

Is that possible for us to calculate an energy emitted from the ROI of a thermal image based on image processing? Let say the above picture, if that box is our ROI, can we calculate the energy or something like temperature level on that segmented region so that we can differentiate between the highest and the lower temperature region on the thermal image?

Regards, Desperately need your advice, [e-mail]

Answer 2

[Mohammed Jaddoa]

Bro I have worked on infrared thermal images , first, try use FLIR software it comes up with library can be used with Matlab ,First use video or image captured by infrared thermal camera I used with seq format in this case you will get an image with gray format and each pixel represent temperature, afterwards you can be used a color map to show color but it is not necessary... most important use raw image from camera so you can get temperature, and you below code to show image with color figure(n);imshow(image,[32 38]);colormap('hot') I hope this comment answer your question