Polygonal shape recognition using string-matching techniques

In this paper we study several aspects of the use of string-matching techniques as an approach to the problem of recognizing and classifying polygons. Several authors have already proposed methods for polygon recognition that are based on string-matching. In many cases, however, linear strings are used to represent polygons, which makes it difficult to handle different orientations of an object efficiently. We can, however, easily extend the linear string-matching techniques to cyclic strings, at some small computational cost. We will propose a method to represent polygons as cyclic strings and we will show how cyclic string-matching techniques can be used for rotation-, translation- and scale-independent polygonal shape recognition. We will, however, also point out the limitations of such an approach.