| 摘要 |
PURPOSE:To recognize characters with high stability and high accuracy, by deciding the dividing style of the character frame in accordance with the degree of complicacy of horizontal and vertical subpatterns of an input character pattern to extract the feature element and referring to a dictionary corresponding to the degree of complicacy. CONSTITUTION:A subpattern extracting part extracts subpatterns HSP, VSP, RSP and LSP showing strokes of horizontal, vertical, right slant and left slant directions respectively on the basis of an input character pattern and the width of lines. A complicacy degree calculating part receives the subpatterns HSP and VSP, the size of the character frame and the width of character lines and counts the black bits within the character frame of each subpattern. Then the sum of black bits is obtained, and the quantities corresponding to the character line density are obtained for both HSP and VSP in the form of the complicacy degrees of the HSP and VSP respectively. Further, an identifying part measures the resemblance between the input character pattern and a standard character mask and then recognizes the character code of the most resemblant standard character mask as an input character pattern name. |
