Contour Filling

Abstract

One of the most common problems in graphics and picture analysis is finding the interior of a region when its contour is given, i.e., transforming a class 3 picture into a class 2 picture. For example, any shading algorithm presumes the solution of this problem. In pattern recognition, many algorithms compute integrals over the area of a region and require knowledge of the interior. In photoScannedting, fonts are often described in terms of contours which are then filled to produce the final copy. The problem can be solved in many ways, which can be divided into two broad classes. In the first, one has a precise description of the contour as a polygon and decides which parts of the plane lie in the interior by considering, in effect, the line equations. Such techniques could be called polygon based but we prefer the shorter term edge filling and describe them in Section 8.2. Methods of the second class map the contour onto the discrete plane and then locate the interior by examining the values of pixels. Such pixel based techniques are discussed in Sections 8.3 to 8.5.