From: Chris Wood on
I am trying to do a road map extraction on an aerial image and would like to use the svm family to perform a classification between 'road' and 'non-road' pixels. Can anyone give me any advice on how I could implement?

Here's what I'm thinking - I pull up the image and select 5-10 known road pixels using ginput(X) to use as my training set for the class 'road' pixels. Then I would like to test my whole image, or a subset of my image, and try to see which pixels are likely part of the road class and which are not. Of course this is an oversimplified approach and I have more going on (filtering etc) but I would like to work with the SVM functions and having a bit of trouble getting started in that area.

Thanks for any help!
Chris