Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

I have an image (can be 2D or 3D), and compute a 2D histogram of the image (for example, the pixel intensity and gradient along certain direction). There is a known target region $R^*$ in the image. I want to find a region $R$ that best matches $R^*$, by defining a convex polygon $h$ in the histogram. $R$ will be the pixels in the image space that are chosen by $h$. my question is:

1) How to optimize the vertices of the polygon such that $R$ best matches $R^*$?

2) If I also want to know the number of vertices, is there any method for that (supposed to be a tradeoff between computation time and the precision, since infinite number of vertices would give a arbitrary convex shape).

As an initial effort, I think the set of points in R is only function of the vertices of $h$, instead of all the interior points of $h$. If I move each vertex one by one, I probably only get local optimal solution. This is probably not a problem that can be solved in polynomial time? I'm not sure if it's related to any 'convex problem'. The functional form $R = f(h)$ seems different from the convex function I know.

share|improve this question
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.