Properties of contour codes

Objects within a binary image are efficiently encoded by traversing their outlines to produce a string of codes each of which defines a directed line to the next point on the outline. Most commonly four or eight codes are defined because these are all that are necessary to define a move from one pixel to an adjacent pixel on a square grid. Many features of the binary objects or blobs may be extracted directly from the code string which describes the outline or contour. The paper draws attention to the fundamental nature of the crack coding scheme, shows that the chain code and midcrack code may be derived from the crack code, and proves relationships between simple measures of perimeter and area as obtained from the different codes and from bit quads. An enclosing octagon is introduced