Search website

Quick links

↑ ↓ to select · Enter to open · Esc to close

Blog

PDF OCR and Highlights: Make Scanned Text Selectable

Updated · 14 min read

PDF OCR and Highlights: Make Scanned Text Selectable

Listen to this post

0:000:00

Share this post

XLinkedInReddit
Contents

If you can't highlight text in a PDF, the file is almost certainly a picture of a page, not a page of text. Running OCR (optical character recognition) adds a hidden text layer on top of that picture. Once it's there, you can select words, highlight them, search them and export them like any normal document.

This guide shows how to confirm that's your problem, how to fix it, what to check afterward, and how to turn highlights into something you will actually reuse.

TL;DR

  • A scanned PDF stores images, so your cursor has no words to grab. A quick drag-to-select test tells you in seconds.
  • OCR adds an invisible, searchable text layer over the original scan, so the page looks the same but behaves like text.
  • You can still draw colored shapes on a non-OCR scan, but those marks are visual only. They can't be searched or exported as clean text.
  • After OCR, spot-check the output, keep to two to four highlight colors, and save your notes next to the passages they refer to.
  • Omphalis is built for the step after OCR: reading, annotating and asking questions of a document you already have in text form.

Table of Contents

Why you can't highlight your PDF

PDFs come in two broad kinds. A PDF exported from Word or a design tool contains real characters. A PDF made by a scanner, a phone camera or a photocopier contains an image of each page. To your eyes they look identical. To your PDF viewer they are completely different.

Indiana University's library guide on accessible research articles describes the image-scan case as common for older journal articles, and points to the same fix: get a real text layer onto the file. Older journal articles, scanned book chapters, a professor's photocopied reading packet and a photographed handout all tend to fall into this group.

The symptoms are consistent:

  • The text cursor never appears when you hover over a line.
  • Click-and-drag selects the whole page as one block, or nothing at all.
  • Ctrl+F (Cmd+F on a Mac) finds nothing, even for a word you can see.
  • Copying and pasting gives you blank space or garbage characters.

How to check if a PDF already has OCR

Don't run OCR blindly. Some files already have a text layer, and redoing it can sometimes make things worse. The check takes under a minute.

  1. Open the file in any viewer, such as your browser, Preview or Acrobat.
  2. Press Ctrl+A (Cmd+A on a Mac) to select all, or click and drag across a paragraph.
  3. If the text lights up line by line, you have a text layer. If nothing is captured, the file is currently just images. The Indiana University guide recommends essentially this test.
  4. Search for a word you can see on the page. If it isn't found, the text layer is missing or unreliable.

A third case is worth knowing about: a PDF that has OCR, but poor OCR. Words highlight, yet the highlight lands a line off, or search finds some words and misses others. That is a recognition-quality problem, covered below.

What OCR actually does to your file

OCR doesn't replace the scan. It reads the image, works out which characters are on it, and then places an invisible text layer in the matching positions over the original picture. According to Adobe's OCR guidance, major PDF editors offer built-in OCR that cleans scanned pages before producing the searchable, selectable layer. Cleaning typically includes straightening a crooked page (deskewing) and removing specks and noise, because a cleaner image gives more accurate recognition.

The layer works because the software stores where each word sits on the page. The Library of Congress technical guidelines for its newspaper digitization program describe OCR text with associated bounding boxes for words, which is what lets a viewer paint a highlight right over the right pixels.

OCR text layer aligned with page word boxes

That's why your scan still looks like the original page: a photograph of the paper with an invisible text layer on top.

How to make PDF text selectable, step by step

The exact buttons vary by program, but the workflow is the same everywhere.

1. Prepare the scan

Garbage in, garbage out. Before you run OCR, check these:

  • Is the page straight, with no heavy shadow along the spine?
  • Is the text sharp enough to read comfortably at full size?
  • Are pages in the right order and the right orientation?

If you're scanning yourself, a flat, evenly lit page gives better results than a quick angled photo. If you can't rescan, rely on the cleanup step your OCR tool offers.

2. Run OCR

Pick any reputable OCR tool: a desktop PDF editor, an online converter, or the scanning app on your phone or copier. Adobe's walkthrough shows the typical flow: upload the PDF, choose the language, let the tool recognize the text, then download the searchable result.

Two practical points:

  • Choose the right language. Recognition engines lean on language models. A French article processed as English will have more errors.
  • Be thoughtful with sensitive files. If the document is confidential, such as medical records, unpublished research or student records, check how the tool handles uploads before using a free online converter. A local desktop tool keeps the file on your machine.

3. Verify the result

Repeat the select-all and search tests from earlier. Then do a harder check: pick three passages, including one with a number, one with a proper name and one near the page edge. Select each and paste it into a blank note. If the pasted text matches the page, the layer is trustworthy.

Common trouble spots are:

  • Footnote numbers and superscripts
  • Tables and multi-column layouts, where reading order can scramble
  • Equations, special symbols and non-English characters
  • Faded or handwritten margin notes

OCR output that merges columns or scrambles reading order will still highlight, but it will read badly if you use text-to-speech or export your highlights.

Highlighting before OCR versus after

You may have noticed some apps let you highlight a scan without running OCR. What they do is draw a translucent rectangle over part of an image. It looks like highlighting, but it's closer to using a marker on a photograph.

Visual markup on a scan Highlight on an OCR text layer
What's stored A colored shape at a position A selection tied to real words
Search your highlights No Yes
Export highlighted text No, only a flattened image Yes, as clean text
Works with read-aloud tools No Yes
Accuracy of selection You aim by eye Snaps to words and lines

The Chrome help documentation on scanned PDFs makes the same distinction: text-based highlighting is persistent and searchable, while markup on an unrecognized scan is only visual. If you plan to come back to this material for an exam, a literature review or a report, take the extra minute to run OCR first.

Where your highlights actually live

When you highlight a PDF, the program doesn't rewrite the page. A highlight is stored as a separate annotation object. Documentation for an open-source PDF editor library explains that modern editors append new annotations to the end of the file, an approach that avoids breaking any digital signatures already in the document.

This has useful consequences:

  • Your highlights travel with the file, but only if the app you used saves them into the PDF itself. Some apps keep highlights in their own database instead.
  • Another viewer may display your highlights differently, or not at all, if the annotations were not written in a standard way.
  • Highlighting never improves OCR. If the text layer is wrong, the highlight will still sit on the right pixels but capture the wrong characters when you copy them.

Before you invest hours in a big reading project, test it: highlight a sentence, save, close the file, and reopen it in a second viewer. If the highlight is still there, you're safe.

Searchable is not the same as accessible

OCR solves the selection problem. It doesn't solve everything. The Library of Congress notes that machine-readable text is only a starting point. For real reuse, a document needs structure such as logical reading order, headings and tagging, which go beyond character recognition.

It also matters for people. The Library of Congress's National Library Service describes how readable text enables screen readers, text-to-speech and synchronized speech and text highlighting for readers with visual impairments or reading disabilities. If you prepare readings for a class or a study group, running OCR is a kindness to classmates who rely on those tools. A readability check on the text you produce, such as this Flesch-Kincaid readability score checker, can help you judge how approachable a passage is once it's in plain text.

A highlighting routine that survives the semester

Once your text is selectable, the temptation is to highlight everything. Don't. A page that's mostly yellow is a page with nothing marked. Here's a routine that holds up.

  1. Read a section before you mark anything. Highlight only after you've finished a paragraph, so you choose what mattered instead of what looked important in the moment.
  2. Use two to four colors at most. Smallpdf's blog suggests limiting highlights to key passages and using a small number of colors, because complicated color systems slow review. A simple scheme works: one color for claims, one for evidence, one for terms you need to define, and one for questions.
  3. Write a note on every highlight that matters. A highlight says "this was important." A note says why. Even six words, such as "contradicts the 2019 study," will save you from rereading.
  4. Export or review within 48 hours. Go back to your marked passages while the context is fresh and pull the best ones into your notes.
  5. Connect the material to a task. If you're reading for a course, line the readings up against your deadlines. A syllabus-to-calendar guide such as this walkthrough on extracting assignments from a syllabus can help you decide which readings to take apart carefully and which to skim.

For a deeper look at turning marked passages into study material, see our guide to a PDF highlighting strategy that turns highlights into flashcards. If you're working on a tablet, our piece on annotating PDFs on iPad covers practical workflows, and the article on highlighting PDFs in Chrome, Edge and Firefox covers browser-based options.

Our perspective: highlights are where reading starts, not where it ends

We build Omphalis for people who save more than they read: students with folders of papers, professionals with a backlog of industry reports, listeners who bookmark podcast episodes they never revisit. OCR is a very common first obstacle for those people, and it's worth solving. But we've seen that fixing the text layer only unlocks the next problem. You now have a document you can mark up, and you still have to understand it.

That's why we think about highlights differently. A pile of yellow lines is a record of what looked important, not proof that you understood it. Our approach is to keep you inside the full source, explain difficult passages right where they appear, and attach your own notes to the original wording so context never gets lost. We would rather help you understand one paper properly than give you a fast summary that leaves you sounding fluent without actually knowing the material.

So our advice is practical. Get a clean text layer. Check it. Highlight sparingly. Then add a note or a question to each mark, so that when you return in three weeks, the passage tells you what you were thinking.

— Omphalis Team

Taking OCR'd readings further with Omphalis

Once your PDF has selectable text, Omphalis gives you a place to read it properly. You can annotate passages with notes tied to the original text, get explanations of complex sections without leaving the page, ask questions and get answers with citations from the source, and listen to content with natural voice narration. It also supports importing PDFs, EPUBs and Word documents, and exporting what you've collected.

There is a Free plan to start with. Paid plans are Pro at 9 USD per month or 90 USD per year, Premium at 19 USD per month or 190 USD per year, and Scholar at 39 USD per month or 390 USD per year. Paid plans add features such as PDF reading, word-by-word read-along and Linked sources. Export is free on every plan. Unit packs start from 10 USD one-off. Full details are on the Omphalis pricing page.

If you read mainly academic papers, take a look at how we handle reading and understanding research papers.

Fixing common OCR and highlight problems

Highlights land on the wrong line. The text layer is misaligned with the image, often because the scan was skewed. Re-scan the page straight, or re-run OCR with cleanup and deskew turned on.

Words are misspelled when you copy them. Recognition errors are most common in small type, faded print and unusual fonts. Try a higher-resolution scan. For a short passage you plan to quote, retype it from the image.

Search finds some words but not others. The OCR language setting may be wrong, or the page has mixed languages. Rerun with the correct language.

Columns read out of order. Multi-column journal layouts can confuse the reading order. Highlighting still works, but export and read-aloud will jumble sentences. Check one two-column page before you process a whole book.

The file got huge after OCR. Some tools save a high-resolution copy of the page images with the text layer. Look for a compression or optimization option in the save settings.

Fixing common OCR and highlight problems — overview diagram

FAQ

How do I make PDF text selectable?

Run OCR on the file using a PDF editor, an online OCR converter or a scanning app. The tool reads the page images and adds a hidden text layer, as described in Adobe's OCR instructions. Afterward, test by selecting a paragraph and pasting it somewhere.

How can I tell if a PDF is OCR'd?

Try to select a line of text by clicking and dragging, or press select-all. If the words highlight individually and search works, a text layer is present, which is the check suggested by the Indiana University library guide. If nothing is captured, the PDF is just page images.

Can I highlight a scanned PDF without OCR?

Sometimes, but only as a visual mark, like drawing a colored box over a picture. Those marks can't be searched or exported as text, which is why Chrome's documentation points users to text-based highlighting. For any serious study or research, run OCR first.

How do I save highlighted text from a PDF?

First make sure the PDF has a text layer, then use a viewer or reading tool that can export annotations as text. Highlights are stored as separate annotation objects, so check that the app saves them into the file or lets you export them. Reopen the file in a second viewer to confirm they persist.

Sources