Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

Quantify how blurred is an image

Hello:
I would like to implement an IMAQ-algorithm so that I am able to quantify the - blurriness - of an image. The aim in my application is to accept only the non-blurred images. I saw some references in the literature speaking about edge detections and Sobel filters but cannot find a way to make it complete. Any help would be appreciated.
Chris
0 Kudos
Message 1 of 3
(5,385 Views)
Typically many of our customers use edge detection with variable slope values to see the steepness of the edge's slope. You may want to also look at our auto-focus example that is on the web.

Auto-Focus Example
0 Kudos
Message 2 of 3
(5,385 Views)

IMHO, I think a measure of blurriness is a relative measurement. In other words, you can use measure of stdev of gray values, or the stdev of the gray values after processing (e.g., edge detection), but the value would need to be referenced to some other value. If the sceen doesn't change, then perhaps you can determine what this value is on a well-focused image. After that, you can set some kind of threshold value of your measurement.

 

 

0 Kudos
Message 3 of 3
(4,217 Views)