Method and system for processing images of forms which have irregular construction and/or determining whether characters are interior to a form

Abstract

A method and system for processing images. An input or scanned image is first processed into a plurality of rectangles. Using these rectangles, a rough classification process is performed to determine if the rectangles define character candidates, horizontal ruled line candidates, vertical line candidates, or form candidates. A processing of the form candidates is performed which allows the determination of whether the form has a regular shape such as a rectangular shape or an irregular shape which includes protrusions from a rectangular shape. The form candidates are analyzed in order to determine if they are irregular forms by constructing and analyzing one or more cores which define the form. After an irregular form is determined, a determination is made as to whether character candidates are inside of or outside of the form. If it is desirable to only perform character recognition on character candidates which are outside of the form, a scanning process is performed in order to determine whether the character candidates are inside of or outside of the form.