In optical character recognition, text strings are extracted from images so that it can be edited, formatted, indexed, searched, or translated. Characters should be grouped into text strings before recognition, but the existing methods cannot group c