【論文メモ】Block segmentation and text area extraction of vertically/horizontally written document

論文名

N. Amamoto, S. Torigoe, Y. Hirogaki: Block segmentation and text area extraction of vertically/horizontally written document, Document Analysis and Recognition, 1993., Proceedings of the Second International Conference, 1993.
Block segmentation and text area extraction of vertically/horizontally written document - IEEE Conference Publication

概要

図表やテキストが混在した文書から、OCRの活用に向けて、縦書き/横書きを問わず文書部分を自動抽出する手法を提案。検証では83％の精度（255 of the 309）を示した。

先行研究との差異

• 先行研究
  • Whal et al.[1] proposed the method to extract blocks by smearing process for a document image.
  • Tsujimoto et al.[2] presented a method to extract adjacent connected components as a segment and to integrate them for the extraction of text area, figure etc.
  • The approach to use white spaces as the basis for segmentation was proposed by Baird et al. [3].
• In this paper, we present the block segmentation and text area extraction method using the white spaces of the document image, without qualifying the form of document such as vertical/horizontal writing.

キモとなる技術や手法

• 2.3 Judgment of Vertical/Horizontal Writing に書かれている
• 先行研究の技術を、日本語に対応できるよう工夫

Block抽出
→Textか判定
→Blockの縦横比で、縦書きか横書きか判定
→全てのBlockで縦書き/横書きのどちらが多いか合計を比較し、文書全体が縦書きか横書きかを判定

有用性の検証

• 45の文書で検証。309あるText部のうち、255を抽出できた
• 平均処理時間は1文書当たり5秒（で、有用といえる）

議論

• 縦書き/横書きが1枚に混在している文書
• 歪んだ入力画像
• 文書全体ではなく部分画像

でも取り組みたい

次に読む

[1] F.M.W ahl, K.Y.Wong, R.G.Casey, "Block segmentation in mixed text/image documents", Computer Graphics and Image Processing, vol.20, pp.375-390, 1982.
[2] S.Tsujimoto and H.Asada, "Major components of a complete text reading system" ,in Proc.of the IEEE, vol.80, No.7, pp.1133-1149, July 1992.
[3] H.S.Baird, "Anatomy of a versatile page reader", in Proc. of the IEEE, vol.80, No.7, pp.1059-1065, July 1992.