| 摘要 |
A computer-implemented method assigns value codes to characters, numbers, special signs, words and the like at code speed for storage and retrieval. Value codes of characters and character groups on a document are produced by scanning the document to make a bit map image of printed material in the document, the bit map being stored in a memory. Lines 1-8 of bits in the memory are analyzed and compared with stored value codes in a character and word structure code table. When matches are found, value codes for the document are stored in a separate memory along with standard computer codes (such as ASCII) and information usable to retrieve the material. The value codes are constructed from all portions of the bit map image of the characters and groups and thereby represent characteristics of the images themselves. After value code determination, i.e., conversion, they can be checked against a code dictionary. Retrieval of stored value codes for documents is accomplished by entering search words in a standard code, locating the value code corresponding to the standard code and conducting a comparison search in the value code. Apparatus for performing the method is disclosed. By the assignment of pixel-related code values, words are automatically converted. Fonts are correlated with logos limiting font selection of assigned value codes. The system can handle already established system fonts or new ones encountered, including those found difficult by prior art OCRs. The method accounts for variations in pixel density and pixel splatter and can handle material in Arabic, Chinese and Japanese. Editing requirements are substantially reduced. <MATH> |
