The reverse would be to look at every sample you have and decides its colors based on the line information (the two points) with basic geometric calculus (distance of a point to a line). Applying anti-aliasing / thicks borders is more easy in this way also.
The reverse is called Backward Drawing / Mapping, and is fit for GPU programming and others uses. Some problems are only defined for one kind of drawing and it's difficult to transform it to the other kind. It's like finding the inverse function of a mathematical function, hard.