| 摘要 |
<p>The object of the present invention provides apparatus and methods for recognizing characters which have been scanned and converted into digitized format wherein a first plurality of curve following scans are performed which develop a plurality of unique binary encoded features dependent on the curve segment encountered. The binary encoded features generated from the performed scans are matched against a plurality of binary encoded patterns of standard canonic curve segments and comparative information about the performed scans is determined relative to the standard canonic curve segments scans dependent upon the shapes of the scans, the starting points of the scans, and angular displacements and lengths of the scans. A plurality of high information data words are accumulated characterizing each of the plurality of performed scans for each character, and a final correlation operation is performed to determine which of a set of standard characters the scanned character represents.</p> |
