| 摘要 |
<p>The method and apparatus disclosed involve the scanning of a representation of a character by means of at least three frame scans based on respectively different directions, the frame scans developing quasitopological codes (QTC) indicative of character STARTS, OPENINGS, CLOSINGS and FINISHES. The QTC relating to convexity-shapes and concavity-shapes of different segments of the character are then automatically stored in respective groups of storage locations in a store, such that, at the end of the process, the groups storing QTC are functionally arranged approximately in the shape of the character so that the character is at least partially recognizable thereby. Also disclosed is method and apparatus in which at least partial recognition of the character takes place by comparing the distances between selected pairs of QTC. In addition, the recognition process may involve the actual comparison of the list of QTC for the character with a set of lists representing predetermined characters.</p> |
