Blog · Accessibility
Dyslexia and reading text in images — when OCR tools help
Every accessibility tool for dyslexia — font overrides, spacing adjustments, colour tints — works on the same raw material: selectable HTML text. The moment that text is baked into an image, all of those tools go silent. The words are still there on screen. But to your browser, they are pixels, not characters, and there is nothing to change.
This is a narrower problem than it might sound, but it surfaces in specific situations that many dyslexic readers find genuinely frustrating: infographics shared on social media, scanned documents saved as PDFs, screenshots of error messages, e-learning slides exported as image files, and product packaging photographed in an online shop. The workarounds are less elegant than applying a font override, but they are surprisingly practical once you know which tool to reach for.
Why image text is a different kind of hard
When dyslexic readers describe what makes text difficult, they usually talk about letter shapes, spacing, and line length. All of those things are typographic properties — and all of them can be addressed in a browser by changing the underlying CSS. A browser font override works by telling Chrome to ignore the site's chosen typeface and render text in something more readable. Letter and line spacing can be adjusted similarly. The text itself stays the same; only the visual presentation changes.
Image text cannot be changed by any of those routes. A JPEG of a printed timetable is a grid of coloured pixels. The letters in it look like letters to a human eye, but they have no encoding that a browser can parse or reformat. There is no CSS hook, no selectable character, no node in the document tree. Whatever contrast, font size, leading, and letter spacing the original designer chose — or the scanner's camera captured — is what you get.
For dyslexic readers this means several compounding problems:
The font is fixed. You cannot switch a scanned legal document from Times New Roman to Atkinson Hyperlegible. Whatever letterforms the original used — and older documents routinely used high-serif typefaces designed for print reproduction, not screen legibility — you read what you see.
The size is fixed in a different sense than normal text. You can zoom in on an image, but because the source is rasterised, zooming large enough to make text comfortable typically introduces blurriness that makes letter shapes harder to distinguish, not easier.
The contrast may be poor. Scanned documents often pick up background texture, yellowing, or faint grid lines that compete visually with the letterforms. Photographs of printed material may have shadows or reflections. These forms of visual noise are particularly disruptive for readers with contrast sensitivity difficulties.
The line length is uncontrollable. A comfortable reading line for most dyslexic adults is roughly 55-75 characters. An infographic designed for landscape social sharing might have lines that run far wider, or a scanned column that is awkwardly narrow.
None of this is fixable at the display layer. The only route to more accessible image text is extraction: get the characters out of the image and back into a form that can be reflowed, restyled, and read aloud.
What OCR actually is
OCR stands for optical character recognition. It is the process of analysing an image — pixel by pixel — and identifying which shapes correspond to which characters in a known alphabet. Modern OCR is accurate to well over 99% on clean, high-contrast printed text, dropping meaningfully on handwriting, degraded scans, unusual typefaces, and text photographed at an angle.
The output of OCR is plain text. Once you have plain text, you are back in familiar territory: you can paste it into a document, a text-to-speech tool, or a plain editing area — and your usual reading accessibility setup kicks in immediately.
The practical question is not whether OCR works (it does, reliably, on most printed material) but which tool gets you from image to text with the least friction.
Google Lens — the fastest route in Chrome
If you use Chrome on a desktop computer, Google Lens is built into the right-click menu and requires no installation. Right-click any image on a webpage and choose "Search image with Google" or "Search image with Lens" depending on your Chrome version. In the panel that opens, click the "Text" tab (it looks like a cursor with lines). Lens will identify all text it can read in the image, and you can select and copy it.
Lens is also available on Android and iOS via the Google app or Google Photos. On mobile, you can photograph a physical page and Lens will extract the text in real time — useful for printed handouts, product packaging, or anything you are holding.
Lens handles most Latin-script printed text well. Where it struggles: low-contrast backgrounds (pale grey on white), ornamental or handwritten fonts, text photographed at more than about 15 degrees off-axis, and very small type below roughly 8pt in the original image.
Chrome's "Read aloud" for images
Chrome has a built-in "Read aloud" feature (accessible from the address bar menu on most pages or via the right-click context menu on some image types). On images with accessible alt text, it will read the alt text. On images without it, it cannot help — but the feature is improving with AI integration and may be able to extract visible text on some image types in newer Chrome versions.
This is less reliable than Lens for image text specifically, but worth checking if you want to stay within the browser's native interface. The underlying text-to-speech is also covered in our guide to text-to-speech tools for Chrome.
AI tools — when the layout matters as much as the text
OCR extracts text, but it does not understand structure. A scanned table might come out as a flat string of cell values with no indication of which column belongs to which header. A two-column infographic might have its left and right columns interleaved. For straightforward prose this does not matter much, but for structured data — tables, timetables, forms — raw OCR output can be harder to parse than the original image.
This is where large-language-model tools (ChatGPT, Claude, Gemini, and similar) add a useful step. You can paste an image directly into any of these tools and ask them to:
Extract the text and present it as readable prose — removing table formatting and column breaks, turning the content into flowing sentences that are easier to read with a dyslexia-friendly font applied.
Summarise the key points — useful when the image contains a dense infographic you need to extract meaning from rather than read word for word.
Reformat a table into plain prose — "The timetable shows that the 08:42 to Edinburgh stops at Carlisle at 10:15 and arrives at Edinburgh Waverley at 11:08" is substantially easier for many dyslexic readers than scanning a grid.
Increase the font and paste into a readable area — once the AI has returned plain text, you can copy it back into any context where your normal accessibility tools apply.
Caveat on accuracy. AI image-reading is impressive but not infallible. On dense tables, complex formatting, or low-quality scans, AI tools occasionally miss rows, swap figures, or paraphrase where they should transcribe verbatim. For any document where accuracy is critical — medical records, legal contracts, financial statements — always verify the extracted text against the original before acting on it.
Dedicated OCR apps and browser extensions
If you regularly deal with image text, a dedicated tool gives more control than right-clicking for Lens each time. A few worth knowing:
Copyfish is a Chrome extension that lets you draw a selection box over any part of the screen and extract the text inside it. It uses Google's OCR engine under the hood and is reliable on standard printed text. Useful for screenshots, images embedded in web pages, and PDF viewers that render pages as images.
Adobe Acrobat's "Recognise text" (OCR) is the standard tool for scanned PDF documents. If you open a scanned PDF in Acrobat and run OCR, the document gains a selectable text layer — which means you can then open it in Chrome and use a dyslexia-friendly PDF reading setup on the extracted text. The free Acrobat Reader does not include this; you need Acrobat Standard or Pro, or an Acrobat online tool.
Microsoft's "Immersive Reader" in Edge can OCR images embedded in PDFs and web pages as part of its accessibility workflow. If you use Edge for work (particularly in Microsoft 365 environments, where Immersive Reader integrates with Teams and OneNote), this can be a zero-install option. It is covered more fully in our guide to Edge for dyslexic readers.
Apple's Live Text (macOS Monterey and later, iOS 15 and later) is the equivalent of Google Lens for Apple devices — point the camera or hover the cursor over an image and select text in real time. If you do most of your reading on a Mac or iPhone, it is the fastest route: no right-click menu, no app to open, just hover and select. Our guide to setting up a Mac for dyslexic readers covers Live Text in more detail.
A practical workflow for common situations
Most image-text problems fall into a handful of recurring categories. Here is the fastest route through each:
Scanned PDF (work document, government form, academic paper): open in Chrome. If the page renders as images rather than selectable text, right-click a page image and use Lens. For multi-page documents, Adobe Acrobat's OCR produces a persistent selectable text layer — worth the extra step if you will return to the document. See also our guide to reading research papers with dyslexia.
Social media infographic or screenshot: right-click in Chrome and use Lens. If the layout is complex (multi-column, heavy on visual hierarchy), paste the Lens output into an AI tool and ask it to render the content as plain prose.
E-learning slides exported as images: this is one of the more frustrating situations, because the images often come inside a learning management system that does not allow right-clicking. Screenshot the relevant slide, open Google Photos or the Google Lens app on your phone, upload the screenshot, and extract the text there. Alternatively, use Copyfish to draw a selection directly over the browser window. See our full piece on dyslexia and online learning for more on LMS accessibility.
Physical printed material (handout, letter, packaging): photograph it with your phone and open in Google Lens or Apple Live Text. Extract the text, copy it, paste into a notes app. Then your usual reading preferences — font, size, colour scheme — apply immediately.
Screenshot of an error message or technical text: paste directly into an AI chat tool and ask it to explain the content in plain language. This sidesteps the transcription step entirely when what you need is comprehension rather than the exact wording. Our guide to reading GitHub with dyslexia covers this scenario for developer contexts.
What LexiFont does — and does not — help with
LexiFont works on live HTML text. On any website, it can swap the font to OpenDyslexic, Lexend, Atkinson Hyperlegible, or Comic Neue; adjust letter spacing and line height; and tint the background colour. For everything that is selectable HTML text — articles, emails, product descriptions, documentation — this works immediately and applies across every site you visit.
For text baked into images, LexiFont cannot help directly. But the two approaches are complementary rather than competing. On a page that mixes regular text with image-embedded charts or infographics, LexiFont handles all of the regular text while you use Lens or an AI tool for the image portions.
If you find yourself dealing with image text often — in an academic or professional role where scanned documents are the norm — the one-time investment of setting up Copyfish in Chrome, and practising the Lens right-click gesture, pays off quickly. LexiFont Pro is the complement for the text side of the same pages: extended font options and granular spacing controls that make the non-image text as comfortable as possible.
Alt text is the right long-term fix. When image text is inaccessible, the root cause is almost always a web content decision: the publisher chose an image where they should have used real HTML text, or they failed to include alt text that describes the image's content. The web accessibility checklist for developers covers when to use images vs text and how to write alt text that actually communicates meaning. If you encounter persistently inaccessible image text on a site you use regularly, feeding that back to the site's accessibility contact is worth doing — many publishers are not aware of the problem until someone flags it.
When the image has good alt text
A final, and often overlooked, route: alt text. If the image was created by someone who followed accessibility guidelines, it will have an alt attribute that describes the content of the image — not just "infographic" or "chart" but the actual data or reading content the image contains. You can inspect this in Chrome by right-clicking the image and choosing "Inspect" — the alt attribute appears in the element code as alt="...".
If the alt text is substantive, you can copy and paste it directly — you now have the content in selectable text form, ready for your reading preferences. Screen readers (covered in our piece on screen readers vs font extensions) also surface alt text automatically, so this route is sometimes the fastest on pages from publishers who invest in accessibility.
In practice, alt text on infographics and charts is rarely complete enough to replace reading the image. But it is worth checking before reaching for Lens.
Further reading
- Dyslexia-friendly PDF reading in Chrome — tools and settings
- Contrast sensitivity and dyslexia — what it means for screen reading
- Reading research papers with dyslexia — a practical workflow
- Dyslexia-friendly web design — a checklist for developers
- Text-to-speech in Chrome for dyslexic readers
- Dyslexia and online learning — navigating LMS platforms