A New Method for Segmenting Newspaper Articles

Digital preservation of old newspapers contributes greatly to the historical register of a country's social, political and economical events. At the same time, newspaper preservation is an imperative necessity because of the fast paper deterioration and difficulty in tracing the overwhelming amount of information. Lambrakis Press S.A. owns a large collection of newspapers and periodicals that consists of 1,300,000 pages and covers a time period from 1890 up to date. This material is divided into 600,000 A2 pages, 500,000 A3 tabloid and 200,000 A4 pages approximately. Our team is working on all aspects of the transformation procedure from the printed material to an accessible digital archive (verification and quality control, digitization, cataloguing, search and retrieval, design and content presentation). The final digital documents form the foundation of our digital library. Preservation and processing of this precious material can be achieved by focusing on a series of problems related to the digitization of the printed material, such as: image enhancement by noise removal, isolation of newspaper articles by document understanding techniques (segmentation-labeling). The successful tackling of these problems allows the subsequent efficient cataloguing by employing OCR, full text retrieval and information extraction techniques along with manual indexing. In our paper we will present the results of our research associated with the stage of segmentation of the various regions-the image consists of-as well as the identification of text regions which have to be separated from other regions, i.e. figures, drawings or line regions. The main region segmentation techniques are based on two fundamental approaches: firstly, on the smearing and labeling of regions [1-2], and secondly on the image profiling in various directions [3-4]. Both techniques have not been successful in achieving newspaper segmentation because of the haphazard lay out of newspaper articles and their very close contact. Furthermore, the first approach results in great computational cost. Aiming at a solution of these particular problems accruing from the newspaper segmentation, we suggest a new technique based on