How can I find a threshold in a point cloud (xyz) separating two classes?

mickey3k

Hi everyone!

I am completely new to RapidMiner and machine learning. I am a student and currently working with point cloud data. That is a text-file representation of a land surface through coordinates. Additionally to xyz coordinates the points have color values in HSI (hue, saturation, intensity) color space. I need to separate vegetation from ground points based on the hue color information. Ground points have a low hue value, vegetation high hue value in this case. So I want to find a threshold that separates those two classes in a representative way.

I know that I can somehow do this with decision trees or support vector machines.. but since I never did something like this I am kind of clueless how to start and how to use those methods.. Is this even possible without further attributes to describe the two classes?

So I would be very pleased if someone could give me hints on how to achieve my goals.

Thanks a ton!

Michael