i have a thermal image taken from flir camera . I want to process it to find the location of the fault and how can we map the location of the fault in the normal image.

i have a thermal image taken from flir camera . I want to process it to find the location of the fault and how can we map the location of the fault in the normal image.

Answer 1

[Image Analyst]

You need to calibrate your camera. Simply knowing some row and column of the pixel where the fault is, is only the first step in the problem (and the easiest step). You then need to know where in real world space the problem is, I guess so you can go out into the field and fix the problem. See https://www.mathworks.com/products/computer-vision/features.html#camera-calibration to learn how to calibrate your camera.

Also, I hope you can get the image in units of degrees. If you're stuck with the pseudocolored image, the best you can do is to try to convert the pseudocolored image into a temperature image by decoding the colormap. See attached demo.

[kameswara]

Hi sir, I have attached the image for ur reference. Kindly suggest me what can I further do with the image and how can I do it

[Image Analyst]

I'm assuming you want to find the location of the fault out there in the real world, like up there on the electrical rig. So first you find the fault in the image. That will give you a pixel row and column, but that doesn't do you much good for your technicians in repairing the fault. They want to know a location like what transformer at what location at your power plant. So to do that you need to turn a row,column coordinate of your image into an actual physical location on your real world site.

So buy the computer vision toolbox. Calibrate your camera. Analyze the image to determine what pixel is faulty. Use the calibration to determine where in your scene in the real world is your faulty point.

[kameswara]

hi sir can u give some code to do the above task

[Image Analyst]

No, I don't have any. The toolboxes usually have demo code though.