0

I am dealing with image processing using Matlab and my goal is to assign pixels to a characteristic line of an object.

In detail:

I take pictures of different metallic objects. I can distinguish the objects in 6 different classes. The classes are known to me only in “my brain”, not in the computer. There is always one class in one picture. In the next step, I use color segmentation to distinguish my object from the background. After the segmentation, I use the Canny operator to find the edges of the object. So far, let’s assume that there is no image “noise” and that the Canny operator determined the contour of the object. This is how far I have come so far.

Two different objects that give you an idea of the "characteristic lines"

In the next step, I want to assign the pixels to its “characteristic lines” which are colored in the given picture. Of course, they are not colored in the edge picture, where I only have the information that a pixel has the value “True” if it is an edge pixel or “false” if a pixel is background.

Do you guys know of an algorithm or method that can assign the edge pixels to “characteristic lines”?

TylerH
  • 20,799
  • 66
  • 75
  • 101
Igor
  • 1
  • 2

1 Answers1

0

Try using MatLab's corner() to break up the edge into sections

http://www.mathworks.com/help/images/ref/corner.html

Andrew Hu
  • 150
  • 1
  • 12
  • Thank you, but I tried this one before and it gives me plenty of corners that are no corners, and the corners that I need are not identified by the corner()-function. This is actually the first step to think of, but did not work well for me, so I asked this forum for ideas and knowledge share. – Igor Feb 09 '16 at 19:53
  • hm sorry I couldn't help you – Andrew Hu Feb 09 '16 at 21:13