Need to use FFT?
I think you can find average diameter by using imfindcircles function, like:
% Read the image and binarize
I = imread('bacteria.jpg');
% Find circles using Hough transform
[centers, radii] = imfindcircles(I,[20 50]);
% Show the result
imshow(I);
viscircles(centers, radii,'EdgeColor','b');
By this script, 426 bacterias were detected, as shown in the following image. And here is an average radius in pixel.
>> mean(radii)
ans =
32.3151
