Procedural approach for content segmentation of old newspaper pages
R.M. Samitha Chathuranga, L. Ranathunga · 2017
This paper presents a sequential and procedural approach for segmentation of old, low quality printed, degraded newspapers in order to extract articles, which is a major step in newspaper digitization. The approach mainly consists of, detection of white space separators, estimation of a dynamic threshold coefficient, extracting away the textual components, detecting line separators, prolonging both white space and line separators, recursive flood filling and segmentation of the page. This research also reveals the importance of consuming Hough Lines Transformation, instead of projection profiles to detect gaps between text lines, text columns and articles. Importance of extracting away the text components before detecting lines in old documents is also highlighted. The results of the research are interestingly competitive with other literature. This emphasizes the importance of collaborative implementation of multiple approaches and the fact that highly successful results are achievable via a well organized sequential procedure of multiple operations for the page segmentation objective.