Document image analysis system

Abstract

Abstract PURPOSE:To easily extract the constitution and the specific area of a document by automatically determining the constituent elements of a document image and the disposition relation among the elements in hierarchical constitution with use of the attribute value of a basic element constituting the document, and an area in which the up-and-down and left-and-right disposition relations are held hierarchically, as guide. CONSTITUTION:A document image is constituted of the elements such as a character shown by a black circle and drawing shown by a square F1 with slant lines. The image is divided into character line areas S1-S7 whose attribute is character, and the picture element descriptive area F1 whose attribute is the picture element. Character information flows from right to left for the character strings S1-S7 in vertical arrangement and from up to down in one character line. Then employing split detection method, master areas T1, T2 are detected, then the right-and-left relation between T1-T2, and the up-and- down relation between the area F1 and than T2 can be detected. Accordingly, the extraction of a character in a character line from the area T1 to T2 or the extraction of the area T2 blow the area F1 can easily be executed.

Inventors

• 👤 Yoshitake Tsuji

Bibliographic Details