In a perfect world, the union of all the thickened polygons would be exactly the input polygon. Can find the difference either left over or leaking out of the input, and compare areas to the input polygon. After major bugs are sorted out, I'd expect the quality to improve as the step size shrinks (by the calculus integration argument -- smaller step size, thinner rectangles to fit the area under the curve)
In a perfect world, the union of all the thickened polygons would be exactly the input polygon. Can find the difference either left over or leaking out of the input, and compare areas to the input polygon. After major bugs are sorted out, I'd expect the quality to improve as the step size shrinks (by the calculus integration argument -- smaller step size, thinner rectangles to fit the area under the curve)