Is this because the floats are flooring to zero when a decimal is returned automatically?
If you divide one integer by another, you get an integer as the result. With what you're doing there's a strong possibility that'll return zero a lot (probably more often than not), or 1, or 2, depending on what the resolution is.If you divide an integer by a float, or a float by an integer, then you get a float back, which is what ScaleImage really requires.
If you try to scale an image by 0 on either axis, then that's when you'll get errors since you're creating an image which is 0 pixels wide/high.