| 摘要 |
PURPOSE:To improve both the sorting efficiency and recognizing speed by extracting the feature points of a 1st type needed for recognition of characters and at the same time extracting the feature points of a 2nd type that are unable to be omitted as long as a relevant character is recognized despite its running hand writing out of those extracted feature points of the 1st type. CONSTITUTION:When the data on a handwritten character pattern is inputted, a 1st type feature point extracting part 11 extracts the feature points of the 1st type out of said character pattern data and supplies them to a 2nd type feature point extracting part 12 and a feature point comparison part 14. The part 12 extracts the feature points of the 2nd type which are unable to be omitted as long as a relevant character is recognized despite its running hand writing out of the input feature points of the 1st type. Then the part 14 compares the feature points of both the 1st and 2nd types of the handwritten character pattern extracted by both parts 11 and 12 with the feature points of the 1st and 2nd types of each dictionary pattern stored in a dictionary pattern storing part 13. In such a way, a candidate dictionary pattern serving as a recognizing candidate is selected through a major sorting job.
|
