how to detect all local minima in radon transform and average distance between them?

I want to detect motion blur parameters in image using radon transform.

im= imread ('blurredimage.tif');
Rd=radon(log(abs(fft2(diff(im)))));
mx=max(Rd(:));
(Row column)=find(mx==Rd)

Here column gives blur angle theta i.e. blur direction.

I got stuck in computing blur length,

Blur length is computed by finding avg dist of local minima in radon transform.

[Walter Roberson]

Is the local minima to be considered along the rows or along the columns? Is one blur length to be calculated overall, or one per row (or per column) ? Could you surf(Rd) and attach it here for us to look at?

[Image Analyst]

Not sure I follow the theory. But anyway, I assum "avg dist" means average distance. But average distance between the location of the local minima(s), and what other point(s)?

By the way, the correct syntax requires brackets not parentheses:

[rows, columns] = find(mx == Rd)

And radon requires 2 input arguments, the image and the angle or array of angles. Why did you not put in any angle(s)?