Abstract
An efficient edge following algorithm is presented. The algorithm is a sequential onepass procedure. Using the average of the merit (mainly edge strength attached to individual point) as the evaluation function, it finds a near optimal edge line between two points on gray scale digitized images. The algorithm can be used iteratively to extract edge line of a region specified roughly by a number of points. It is typically applicable for the edge line drawing of cut mask making. Generally, finding an optimal path between two separate points is to minimize or to maximize the merit function. The merit function is given as the sum of the merit on the path. So it likely finds a shorter path as optimal in minimization or a longer path in maximization. In that sense the average of the merit on the path is a better evaluation function in finding an optimal path. The divisions in the averaging, however, are a drawback. A large number of divisions sacrifice the processing speed. Some consideration is made to avoid the divisions. A number of experiments are done with good results.
