Normalized Cyclic Edit Distances and Applications

G. Peris and A. Marzal (Spain)


Cyclic strings, pattern recognition, normalization, edit distance, cyclic edit distance, OCR.


Cyclic strings are sequences with no beginning or end, which are useful for modelling different objects in pattern recognition. For instance, in digital image analysis, strings can represent contours of 2D objects. Several methods have been described to compute the dissimilarity of two cyclic strings such as the cyclic edit distance. However, no method has been provided to take into account the normal ization of the cyclic edit distance. In this paper, we propose an algorithm to compute normalized cyclic edit distances, and illustrate the performance of the method in practice.

Important Links:

Go Back