Scanned PDFs and OCR: get editable text from image-only pages
A scanned PDF may contain only photographs of pages. You can convert those pages into image-based HTML, but editable text requires an existing text layer or an OCR step. PDF2WEB exports page images and extracts existing text; it does not perform OCR.
Check whether text already exists
In a PDF tool that supports selection or text search, try finding a distinctive word and copying a sentence. A scan with an OCR layer can provide text even though the page looks like an image. A failed search is a useful clue, not a complete diagnosis: fonts, language and damaged encoding can also affect extraction. You can also run PDF2WEB’s text export and read its notices for pages with no selectable text.
Keep the page appearance
- Open the HTML converter and select the scanned PDF.
- Choose Page images. This renders the page contents into images inside one HTML file.
- Convert and download the file, then inspect every page at a readable size.
- Check the file size before publishing. Image-heavy exports can be substantially larger than text-only exports.
When you need editable text
Run OCR using a tool that supports the document’s language, then export a PDF with a searchable text layer. Check that OCR correctly recognized names, dates, columns and numerical values. Feed that PDF into text mode afterward. OCR mistakes are carried into downstream HTML; conversion cannot determine which number or word the author intended.
How do I turn a scanned PDF into Markdown?
Start with one page that contains a heading, a paragraph and a list. Run OCR in the document’s language and proofread the resulting text. Compare similar characters such as O and 0, check minus signs in figures, and follow each column from top to bottom. A searchable PDF can still have an incorrect reading order.
Copy the checked text into a Markdown editor. Put ## before a section heading, start each bullet on its own line with - , and add a blank line between paragraphs. Preview the result next to the scan. Keep a caption and description with any diagram whose meaning the text alone cannot convey. PDF2WEB can extract an existing text layer as HTML for this handoff. OCR and Markdown authoring take place in your chosen tools.
Understand accessibility and search limits
An image of a paragraph is not the same as semantic text. A page-image export alone does not provide equivalent reading access for every assistive technology, and it is not a guarantee of search indexing. Add reviewed text and appropriate structure when publishing an accessible article. See the local conversion workflow and the publishing checklist for the remaining steps.