| 摘要 |
A character reading system which reduces the amount of correction and verification work needed for a document for which characters have been correctly recognized. Whole image data WIMG of a document read by a remote OCR installed at a local station and first character data DATA1 recognized on the basis of a first character recognition method from this data WIMG are stored via a communication network in a memory component provided to a central station. A second recognition component provided to the central station recognizes second character data on the basis of a second character recognition method that is different from the first character recognition method from the image data WIMG read out from the memory component. A decision component decides whether there is a match between these first and second sets of character data. If there is a match, the data is outputted as correct character data, that is, third character data, to a host computer. If there is a mismatch, the data is changed to correct character data by a correction component, and, if needed, this correction is verified, after which the data is outputted as corrected correct character data to the host computer.
|
