0

I have an input as a 3D binary image and the preferred output below:

Input:

INPUT

Preferred Output:

OUTPUT

What image processing methods should I look for if I am to have only the spiky object(s) remain, just like the preferred output above?

Karl
  • 5,613
  • 13
  • 73
  • 107
  • If you cross-post to multiple StackExchange sites, please at least link to the other instances of the question: http://dsp.stackexchange.com/questions/1617/image-processing-method-for-spiky-selection – Niki Mar 03 '12 at 11:04

1 Answers1

1

Well, first of all i'd try to remove zall those white line and get only the pattern. You image seem way to erratic for me to work with. If your pattern is quite regular (as it seems here), you can search for a template and use a correlation to extract interesting parts of the image.

After this, only the 4 big patterns should be left. I would perform some calculations (mainly shape based), some examples here : http://opencv.willowgarage.com/documentation/cpp/imgproc_structural_analysis_and_shape_descriptors.html) I'm sure there are some simple descriptors for this. I think of energy, perimeter, these kind of things.

jlengrand
  • 12,152
  • 14
  • 57
  • 87