Image region segmentation method, and image binarization method

摘要

PURPOSE: To extract a character area out of various input images without performing any complicated processing and to exactly divide the area correspond ing to the attribute of an image by dividing the image area while defining proximate pixels, for which density level difference is less than a threshold value, as the same area. ;CONSTITUTION: As the image information of the image read by a scanner 1, a position coordinate and a density level for each picture-element (pixel) are stored in an image buffer 2, for example. A relation discriminating part 3 performs labelling processing for applying a common label to two pixels, for which the position coordinates are mutually proximate and the difference of density levels is smaller than the threshold from the image information stored in the image buffer 2. A divided result output part 5 divides the pixels in the image, for which the value of an element in a correspondent label buffer 4 is common, into the area composed of pixels having the same label. An area attribute discriminating part 7 discriminates the attributes of respective divided areas based on the geometrical features of respective areas, density distribution and mutual inclusive relationship.;COPYRIGHT: (C)1997,JPO