Subpixel Edge Location in Binary Images Using Dithering

This paper concerns the problem of obtaining subpixel estimates of the locations of straight lines in digital images for purposes of machine vision. In particular, it presents a dithering method for improving the estimation accuracy on a rectangular sampling lattice. By adding uniformly distributed independent random noise it is shown that estimation bias may be removed and that the estimation variance is inversely proportional to the length of the line segment. The sensitivity to incorrect dither amplitude is calculated, and a novel approach is given for adding the dither by using grey-level image sensor and utilizing the imaging model.