1 (edited by Felix 2015-03-07 11:15:18)

Topic: Is there Anything Wrong with the Analyzer's Fourier Transform?

I created a blank, white image (bmp format) with MS Windows Paint , and let Image Analyzer to carry out a Fourier Tansform on it. The output, to my surprise, was a white cross, as shown in the first picture below, whereas the expected result should be as the second picture shows (a white dot in the centre of the frequency domain).

If a constant array is discrete-Fourier-tansformed, the result should be an array containing only one non-zero number, that is, the zero frequency componet, shown in the frequency domain as a white dot in the centre. But the white cross is certainly far from being a dot. Also it does not look as if the big white cross had been left by some acceptable computational errors.

It puzzles me.

http://s13.postimg.org/mz8l1koeb/2015_03_07_110938.jpg

http://s15.postimg.org/yy1ebdk9z/66_ft.jpg

This third image is what was used for the transform (the image hosting website has changed its format from bmp to png):

http://s11.postimg.org/utdi3hilr/image.jpg

2

Re: Is there Anything Wrong with the Analyzer's Fourier Transform?

I don't remember the details of the implementation, but I think that the image is zero-padded up to the nearest power of 2 to be able to run a FFT. This means that you are actually seeing the Fourier transform of a step function.
Try to make your image 256x256 and see what happens then.

Michael Vinther

> software developer <

Re: Is there Anything Wrong with the Analyzer's Fourier Transform?

Thank you for the answer! My puzzle is gone. I resized the image to 256x256 and got as its FT a centred white dot.  : )