Method for extracting structure of document picture

Abstract

Abstract PURPOSE:To extract the structure of a document picture at a high speed and with high stability by deleting the vertical and horizontal segments out of the subject document picture, dividing a white part by rectangles and applying a coupled labeling process to the non-white part. CONSTITUTION:In step 1 the long segment elements (vertical and horizontal components only) of a subject document picture are extracted and erased. In step 2 the mesh division is given to the picture. In step 3 a white area is divided by rectangles. In step 4 a labeling process is applied to a non-white rectangular area and therefore the character rows and pattern parts are segmented in rectangles. In step 5 the non-white areas are united and the area of the document picture is extracted. In such a way, the non-white area can be segmented as a large block and therefore the structure of the document picture can be extracted at a high speed and with high stability.

Inventors

• 👤 Koichi Ejiri
• 👤 公一 江尻
• 👤 Akira Sakurai
• 👤 彰 桜井

Bibliographic Details