Process Design to Self-extract Text from Images for Similarity Check

Abstract

In professional writing, plagiarism is an offensive fraud and a breach of academic ethics. While similarity check is one of the preliminary stages toward an effort to restrain the plagiarism in the domain, most of them fail to detect the text present over the images. The loophole can be used to put the plagiarized text in the form of an image to bypass the similarity check software. To overcome the problem, here, we propose an approach, supplement to the existing software, to extract machine-readable text from the images in scientific documents. The approach accepts portable document format (PDF) and uses the metadata retrieved from the document to automatically detect and localize the images in the document. Thereafter, the electronic form of the text is extracted from the images using optical character recognition (OCR) technique. The proposed framework is validated using a scientific article format containing a block of text and flow diagrams in the form of images. The results indicate that the proposed method is capable of detecting and localizing the text correctly present over the images in machine-readable form.