Distance Transformations are usually defined for binary images, and therefore subject to sampling noise. This paper presents a distance correction that can be applied to any existing distance transform algorithm to achieve sub-pixel accuracy, and evaluates the result. Measurements display a significant improvement in precision.