Contour reconstruction for scaling of digital binary images

Binary images consist of a grid of square cells, which are colored either black or white depending on whether the center of the cell lies in this black or white region of the pre-image. Equivalently, the digitized boundary can be interpreted as a 4-directional chain code of the edge contour. We consider subpixel accuracy in determining the boundary of the pre-image and give a practical algorithm for its implementation. In particular we consider small curving objects digitized on a coarse grid. Experimental results show a high degree of reconstruction, allowing for the upward scaling of small objects from their digitized representation.