| 摘要 |
PROBLEM TO BE SOLVED: To realize a processing speed equivalent to that of a 'frame recognition' method by a 'no-frame recognition' method with high recognition precision. SOLUTION: A character segmentation part 2 executes segmentation processing for each stroke inputted from a stroke input device 1 together with an already inputted stroke in each input of a stroke. A character recognition part 3 executes character recognition processing in each segmentation of a new character. A language processing part 4 executes work dictionary collation processing by combining each character recognition result candidate for a character with an already obtained character recognition result candidate, whenever one character recognition result candidate is acquired. The segmentation processing, character recognition processing and word dictionary collation processing are executed, by utilizing time held up to the start of input of a succeeding stroke. Since it is unnecessary to execute the word dictionary collation processing by post-processing after ending the character recognition processing for all input characters by 'no-frame recognition', a processing speed equivalent to that of the 'frame recognition' method can be materialized with high recognition precision. |
