how to quantify edge roughness

[cloudia0011]

Hi, there,

I am new to the field of imaging processing. This may seem to be very trival to human eye and maybe very trival to people in this field as well, but I need soem help. I am trying to classify objects with similar shape, soem of them have smooth edges and some of them have rather rought edge. I wonder how I can quantify the edge roughness for the comoputer? How do I compute or define it when I have the 2D images with the object edges known? Like the two object in the attached picture.

Thanks for any help!

cloudia

 

[BradtheRad]

Of course you have considered a few approaches already?

One way is to think of the shape as an island. Then you calculate how many miles of shoreline it has, relative to its area. The more shoreline, the greater the roughness.

On this basis your shape at left scores low. The one at right scores high.

However this does not work for a narrow rectangle, which scores high yet its outline is not what you would call rough.

It seems that roughness is not an easy factor to quantify. When you think about it, the scale will be an arbitrary one. A perfect circle should score zero (or 1 depending on where you want to start). A circular saw disk might score 100.

However consider a fine tooth saw. It starts to resemble a smooth circle again. Therefore we should set a threshold as to how fine the scale should be. To put it another way, the scale ought to be proportional to the overall size of the shape.

 

[cloudia0011]

Thanks,BradtheRad! This is very useful information to me!