A Novel Approach for Newspaper Block Segmentation using Run-Length Smoothing Algorithm
Shridevi Soma, Shilpa Shilpa · 2022 IEEE 2nd International Conference on Mobile Networks and Wireless Communications (ICMNWC) · 2022
Region findings and analysis plays an important role in document understanding. Due to the existence of complex layouts, document understanding is a very challenging task for researchers. Layouts in newspapers are derived for the articles where article consists of multiple blocks. These various blocks must be segmented and identified within the whole newspaper which helps further in article segmentation. This paper proposes a novel method to identify and segment blocks found within the newspaper irrespective of its layout using simple image processing operations such as morphological dilation, run-length smoothing algorithm and rule based algorithm. These methods have been tested on dataset consisting of digital enewspaper images of recent years from the TOI and Financial Express Newspapers with different layouts and complexities. The experimental results exhibit our method proposed outperformed region findings with the precision 0.83, recall 0.72 and F1 score 0.76. Effectively these blocks will be used as features and evaluation measure for various document analysis.