DOCUMENT MATCHING METHOD, DESCRIBER GENERATING METHOD, DATA PROCESSING SYSTEM AND STORAGE MEDIUM

摘要

PROBLEM TO BE SOLVED: To attain a highly precise document matching with a little calculation quantity without expanding a document image even when the image is compressed. SOLUTION: On a rough matching stage 15, a bit distribution 25 based on the number of bits required for encoding each scan line is generated from a compressed reference document 12 and the search range of an image database 14 is narrowed down while using a macro statistic quantity 27 calculated from the bit distribution. The bit distribution 25 and the bit distribution of a document in the search range are mutually correlated and candidates 17 are found out. On a detailed matching stage, detailed matching is performed between the reference document 12 and candidates 17 while using end point features.