Transformational Models For Background Subtraction With Moving Cameras