How can you 'quantify' IFFT of 2D image?

Hello,

I have 2D electron microscopy images that can be transform to multiple frequencies using FFT.

I think that the easiest way to analyse these complex images (i.e. spacial domain images) would be to look on multiple frequency domains; thus, I've applied masks on specific freqencies and took the IFFT2 function

Now, I look for a quantitative way to measure how 'imperfect' the image is (see the images attached side by side as an example).

I'm wondering whether there are tools in Matlab that can help me do it?

P.S. I've looked on the different functions under 'image analysis', e.g. entropy, but haven't found something suitable.

Any ideas?

Thanks!

Answer 1

[Alamanda Ponappa Poovaya]

I understand you want to quantify the imperfections in your image compared to your pristine reference image. I suggest you look at the functions in the link below. Particularly, since you have a pristine reference image, the Full-Reference Quality Metrics functions should be appropriate. You can read more about them in the link below

https://www.mathworks.com/help/images/image-quality-metrics.html

[Iron1759]

Thanks...

I actually had some time to dig into the subject further since I wrote this question. The field of image/video quality assessment is a broad field with many intersting approaches.

The full-reference algorithms can't be applied to my images because there isn't a pixel-by-pixel match, unless I align them. The non-reference image quality assessment uses several approaches on spacial, DCT, Wavelet, curvelet, FFT domains etc, and I was inspired by some of these ideas. Now, I'm trying to find something suitable. However, these algorithms based (for the most part) on 'natural scene', and these HRTEM images are far from natural.

Will see what I can get out of it :)