How to extract features from a thermal Image and which is the best technique to extract?

2 views (last 30 days)
I have to detect smallest deviation of temperature in my thermal video(map) by comparing it to the authentic one (training). What are the different attributes that should I extract and which technique should I apply for extraction? e.g PCA, 2DPCA. Based on these features I want to apply machine learning to detect the difference. Thanks

Accepted Answer

Image Analyst
Image Analyst on 18 Jan 2017
Use imregister() to align your new image with your reference image. Then subtract them to determine the temperature difference between the two.
  8 Comments
Qazi  Arbab Ahmed
Qazi Arbab Ahmed on 25 Jan 2017
Thanks, Its helpful. I have 100 thermal Authentic Image and 100 Malicious Images, I want to train them and also want to extract features from them to apply machine learning to distinguish between these classes. What are the features that should be used to distinguish and which machine learning classifier will be effective? Currently I am trying to extract features like mean and SD but I could make response vector, as I am trying classification tool box that requires response and I am not getting it. If you have got idea could you please write some code. Thanks
Image Analyst
Image Analyst on 25 Jan 2017
There's no way to answer this given the information and images you've provided (or not provided). LOOK HERE for ideas.

Sign in to comment.

More Answers (0)

Community Treasure Hunt

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

Start Hunting!