Text line extraction in graphical documents using background and foreground information

摘要: In graphical documents (e.g., maps, engineering drawings), artistic etc., the text lines are annotated in multiple orientations or curvilinear way to illustrate different locations symbols. For optical character recognition of such documents, individual from need be extracted. this paper, we propose a novel method segment and is based on foreground background information components. To effectively utilize information, water reservoir concept used here. proposed scheme, at first, components detected grouped into clusters hierarchical using size positional information. Next, extended two extreme sides determine potential candidate regions. Finally, with help these regions, The experimental results presented datasets camera-based warped noisy images containing seals, etc. demonstrate that our approach robust invariant orientation present document.