Extension of Isovist concept for a point - to Isovist for a polygon

Question

There is the concept of Isovist/Visibility polygon. They both talking about volume of space visible from a given point in space.

My question: What is the algorithamic solution of this problem for a polygon - volume of space visible from a given polygon in space?

I started to thinking:

1. Reduce polygon to line, and then reduce line to point.
2. So there is the solution for a point, then the solution for line is the union of visibility region's (Isovist's) of it's (line) edges.
3. Then the solution for polygon is the union of all visibility region's of it's (polygon) edges.

Is that make sense?

Also asked at stackoverflow