abstract |
(57) [Summary] [Problem] By clustering color information, Provided is a method for extracting characters from a color document image, in which the background color and the character color of an image are separated and characters can be extracted even from complicated and diverse backgrounds. SOLUTION: This is a character extracting method from a color document image in which only a character color portion is extracted from a color document image having a complex and various backgrounds, wherein dither is removed from the color document image by smoothing. From the RGB of the color value of the image from which the dither has been removed, L * u * v * F, perform fuzzy clustering of color information, create a color-separated image (binary image) based on the degree of membership, and remove noise from the binary image. Next, labeling of black pixels and white pixels is performed, and then a binary image suitable for character extraction is selected. Next, character lines are extracted. |