摘要 |
PROBLEM TO BE SOLVED: To retrieve a digitized document image, that is an origin of a document image such as a printed matter, accurately at high speed.SOLUTION: A feature amount detector of document image determines, for each pixel of a document image, the distance between the color information of a base area and the color information of the pixel, and calculates this distance as the concentration signal of the pixel. The feature amount detector calculates the projection concentration of all pixels in at least one of traverse and longitudinal directions of the document image, based on the concentration signal of the pixel. The feature amount detector calculates the concentration distribution statistic in at least one direction, from the projection concentration in at least one direction. The feature amount detector determines a division line for dividing a document image into a plurality of areas, based on the concentration distribution statistic. The feature amount detector determines the concentration ratio feature amount of a document image from the concentration information in each image area, obtained by dividing a document image by the division line. |