How to determine the blurness of an image after using the canny filter?

I would like to get a “blurness” value after using the cannyfilter on a siemensstar image, to check the blurness level. After usind the canny filter the images looks like these:

How can I determine now which image is sharper? Do I need to check the mean value of the histogram or…?


why do this after canny, why not on the source image? why use canny at all? canny is a newbie trap

because only with the canny you can destinguish between “blurry” and “sharp”. After the canny you can see much more detailed lines as in image on the left.

“only”? no.

it’s a fair approach but clearly there’s more going on because the “no lines” area isn’t centered on the Siemens star’s center.

just show your input image. then we can discuss ways to assess sharpness.

ok for sure there are more options, correct. The input image is simply:

So what you suggest?

image is stretched.

that fiducial in the center surprises me. that’s not part of regular Siemenssterne.

the rays appear fairly distinguishable, almost up until the quadrant target’s perimeter.

you should locate the center (that’s probably what the fiducial is there for), then sample the picture in circular 1D paths of various radii. assess each 1D signal for peak-to-peak amplitude.

I’m not inclined to un-stretch that picture you have there. if you can justify its stretched appearance or provide an unharmed one, I might demonstrate my suggestion. make sure the picture is source data. it’s supposed to assess the degradation of one step in an imaging pipeline. any further degradation of the measurement (filtering, resampling, “copying” by taking a snapshot with a phone, … the usual stuff) would defeat the point.

yes it looks a bit stretched but this anyway not the topic. Here a sample of the “blured” image:

So you mention “sample the image in circular 1D path”. Could you explain a bit more on that technic?

What is wrong with the canny filter when the image shows some clear differences?