Character recognition method and apparatus based on 0-1 pattern representation of histogram of character image

摘要

Predetermined recognition units are extracted from an input document image. Histograms are prepared based on the extracted recognition units. By partially collating a 0-1 pattern representation extracted from the prepared histograms of each recognition unit with 0-1 pattern representations of respective recognition target categories, which are prepared in advance, candidates of recognition target categories corresponding to the recognition unit are retrieved.