Segmenting a page of a document into areas which are text and areas which are halftone

Claim 1

An image characteristic identifying method comprising the steps of: a data inputting step comprising scanning a document along scanlines parallel to a main scanning direction; a digitizing step comprising digitizing said scanlines through a bi-level digitization in which a pixel is classified as black or white depending on how its reflectivity compares with a threshold value, and storing said scanlines as bitmaps or run lengths; a detecting step comprising detecting black and white segments along said scanlines; a first linking step comprising linking to each other black segments which are adjacent to each other along a scanline and linking to each other white segments which are adjacent to each other along a scanline; a finding step comprising finding grey intervals along the respective scanlines, wherein each of said grey intervals is bound along a scanline on one side by a relatively long white segment and on the other side either by another relatively long white segment or by an edge of the document, and each of said grey interval comprises either a black segment or a sequence of alternating black and white segments in which adjacent black segments are spaced along a scanline by a relatively short white segment; a second linking step comprising linking to each other adjacent grey intervals into areas of grey intervals where each area of grey intervals comprises grey intervals along a number of scanlines; a storing step comprising storing the positions of said areas of grey intervals and the positions of the black segments and white segments within said areas of grey intervals; and an identifying step comprising identifying such areas of grey intervals on the document as text areas or halftone areas to thereby segment the document into areas which are text and areas which are halftone.