| 摘要 |
PURPOSE:To indentify accurately character without using a large-scale dictionary memory by sorting a character pattern feature into the group of each character and obtaining an average pattern for each group. CONSTITUTION:A feature memory 3 stroes the features of a handwritten character pattern for each character, and a sorting part 4 reads the features out of the memory 3 and sorts them into groups every analogous characters. When a decision processing part 5 decides that the sorting process is through, an average pattern extracting part 6 calculates the average pattern features from the character pattern feature groups given from the part 4. In addition, an identifying part 8 identifies the character to which the average pattern features obtained at the part 6 are equivalent on the basis of a standard pattern stored previously in a dictionary memory 9.
|
