| Proj No. | A1047-251 |
| Title | End-to-End English Newspaper Digitalisation System |
| Summary | Recent advancements in Optical Character Recognition (OCR) technology has led to it being adopted for various tasks such as document management, data entry automation and accessibility tools for the visually impaired. However, OCR of newspaper pages remain difficult, with 90% of historic newspapers remaining undigitised. Currently, there is no open-source OCR engine which can accurately reconstruct the text layout of newspaper clippings as OCR engines struggle to handle multi-column layouts, captions and ads. Since digitised newspapers are valuable for libraries and researchers, there is a strong need for methods that output clean, correctly ordered text rather than fragmented lines. Prior work shows layout parsing helps, but bounding-box segmentation can still break paragraph structure (e.g., lost indentations) and scramble reading order unless columns are explicitly reconstructed and sorted. This project aims to develop a layout-aware OCR pipeline for English newspaper digitisation with post-OCR correction and a user-friendly GUI. The expected outcome is a deployable tool that delivers clean, correctly ordered text with significantly reduced errors for newspaper archival purposes. |
| Supervisor | A/P Ling Keck Voon (Loc:S2 > S2 B2A > S2 B2A 22, Ext: +65 67905567) |
| Co-Supervisor | - |
| RI Co-Supervisor | - |
| Lab | Computer Engineering II (Loc: S2-B3b-08) |
| Single/Group: | Single |
| Area: | Digital Media Processing and Computer Engineering |
| ISP/RI/SMP/SCP?: |