Abstract: A segmentation method and system for detecting intercharacter boundaries for a digital representation of a string of two or more characters extending in a first (X) direction. The digital representation has the form of an m x n array of bits, each being representative of a characteristic (such as optical reflectivity) of a correspondingly positioned picture element (pixel) in an image of the string of characters. A serpentine processor determines a succession of values of interest per swath in the array. A decision network processes the succession of values determined by the serpentine processor to identify a nominal boundary where segmentation is to occur. A separation processor performs the segmentation by associating the various bits representative of black pixels near the nominal intercharacter boundary with either the character to the left or the character to the right of that nominal boundary .

Abstract: A system for processing digitized scan data representative of an image including at least one line of symbols. The system processes the scan data to identify bits including at least one line of symbols representative of horizontal line segments in the image which have a minimum predetermined length in the horizontal direction. Underline tag signals are generated which are representative of those identified horizontal line segments which lie below text characters in the line of symbols being processed. The system further processes the scan data to filter, or remove, all of the bits representative of horizontal line segments to provide a character data signal, where the character data signal is representative of the line of symbols with the predetermined length horizontal lines stripped away. The system then processes the character signals to identify the various text characters, and to generate character signals representative of those identified characters.