METHOD OF DETECTING TEXT IN BITMAP IMAGE (VERSIONS) AND METHOD OF DETECTING SPAM, CONTAINING BITMAP IMAGES Russian patent published in 2009 - IPC G06K9/36 

FIELD: information technology.

SUBSTANCE: invention relates to detection of text in a bitmap image and a method of detecting spam, containing a bitmap image. The method of detecting text involves (a) identification of background colour in a two-colour bitmap image; (b) finding closed contours around each separate image of combined colour pixels, distinct from the said background colour; (c) exception from examination of contours which are too small or too big; (d) finding the presumed alpha characters on all closed contours, interpreted as contours of alpha characters; (e) division of lines of alpha characters found into sets, interpreted as probable words; (f) exception from examination of words in the lines which are too small or too big; (g) counting the number of pixels of non-background colour inside contours remaining after filtration; (h) determining on content of text in the image based on the ratio of the number of pixels of non-background colour inside contours remaining after filtration to the total number of pixels of non-background colour and under the condition that, at least one line remains.

EFFECT: simple and reliable method of detecting text and spam in bitmap images.

17 cl, 57 dwg