以電腦視覺為基礎之圖書數位化自動化流程之研究

Abstract

本研究的圖書數位化工作，使用高拍儀掃描紙本圖書，將拍攝的書頁影像打包成電子書。然而，在數位化的過程中，拍攝的影像往往會有多餘的區域需要裁切，或影像出現歪斜需要糾正。此外，修正完成的書頁圖片需要進行格式轉檔，同時需要透過文字辨識來加速索引數位化。為了達成這些需求，本研究運用了電腦視覺技術，結合影像處理方法，如輪廓查找和影像膨脹，開發了一套自動化輔助處理流程。這套自動化流程大幅提升了工作效率，使效率達到原來的五倍，成功減少了圖書數位化所需的人力需求。

This study focuses on the digitization of books using book scanners to capture images of book pages and compile them into electronic books. However, during the digitization process, the captured images often contain extraneous areas that need to be cropped, or the images may be skewed and require correction. Additionally, the corrected book page images need to be converted to different formats and subjected to optical character recognition (OCR) to expedite index digitization. To meet these requirements, this study employs computer vision techniques combined with image processing methods such as contour detection and image dilation to develop an automated processing workflow. This automated workflow significantly enhances efficiency, increasing it fivefold and substantially reducing the manpower required for book digitization.