Working on a libSVM GUI: how to get classification border?

Hi guys,

I was working on a C++ wrapper and a gui for libsvm. When it's ready it will be opensource.

Right now, I can load data, plot it (if 2d), train a model and highlight the support vectors. Screenshot.

In their example app svm-toy, the libsvm autors iterate over the whole graphic window and for every pixel they call svm_predict to see which class it belongs to, then colorize it appropriately. That works but looks ugly and i don't want to do it.

So my question? How can I find the classifying curve so I can plot it nicely? Any help from more experienced libsvm users would be greatly appreciated.

Thanks.