My question is about edge descriptors produced from an edge detection method applied on an object image. As we know, an edge has four descriptors edge normal, edge direction, edge position, and edge strength. These descriptors are used for differentiating between the detected edges.

Are all of this descriptors are necessary? When can I dismiss any of these descriptors?

I am thinking about assumptions based on the properties of the input image, e.g., if the input image has no noise, can I neglect the edge strength and assume that all resulting edge points are true edges?

  • 1
    $\begingroup$ What are you using the descriptors for? What kind of input images do you have? Necessary for what? Without more info, it's hard to impossible to answer meaningfully $\endgroup$
    – penelope
    May 31, 2012 at 9:56
  • $\begingroup$ @penelope i have edited the question. $\endgroup$
    – HforHesham
    May 31, 2012 at 10:36

1 Answer 1


You can calculate the edge direction from the edge normal by atan2(y,x)link. The other descriptors are independent of each other and thus only the problem itself can tell if they are necessary or not.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.