In automatic symbol recognition, symbols are divided into classes based upon certain characteristics. In order to realize in a concrete manner, a class distribution of pregiven symbols, it is important to locate suitable characteristic parameters of the symbol. The finding of characteristic parameters is considerably facilitated by the fact that for the most part only a relatively slight number of different symbols are to be differentiated. In the present work, a symbol is fixed by means of a matrix that the authors calls the distance matrix. From this, some symbol characteristics that are suitable for classification can be relatively easily read. Author
Edited trans. from Elektronische Informationsverarbeitung und Kybernetik (East Germany) v6 n3 p145-153 1970, by J. Stock.