1
$\begingroup$

Sorry if this question seems to trivial, but I am a bit of a novice when it comes to signal and image processing and I need some guidance.

I have a 3D stack of about 256 grey scale images, each with pixel resolution 256x256, coming from an x-ray tomography scan of a rock sample. I want to segment the data to isolate the void space from the solid space. Typically I use ImageJ software to segment it using its 'auto' feature based upon a histogram of the grey-scale color values. I am wondering if using fourier transform of the grey-scale data can somehow help me in the process. What can the FT tell me about the image data that I can use for segmentation purposes?

$\endgroup$
4
$\begingroup$

Cited from wiki:

"a function (of time) is a representation of a signal with perfect time resolution, but no frequency information, while the Fourier transform has perfect frequency resolution, but no time information."

In your case, the Fourier transform has no spatial information. So for segmentation, you'd better try time-frequency analysis, such as wavelet transform

$\endgroup$
1
$\begingroup$

If I am understanding what you're asking for, you're looking for an algorithm that uses the FT for image segmentation.

1) the FT will tell you where in your image is there a high frequency component. High frequency components represent points where there are large jumps in pixel values, i.e., segment.

2) Segmentation is not done in the FT domain, but rather the image domain. Typically because of 1) above, you will almost always use a high pass filter (such as roberts cross or sobel filter)

3) Given 2), The best thing you can do is look into Canny Edge Detection, it's a major algorithm used in image processing for segmentation. (here is a java implementation of the algorithm)

you can find more information in Digital Image Processing By Gonzalez & Woods

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.