How to Extract Text from an Image and Translate It

You have an image with text in another language. What you do next depends on what you actually need — and most guides skip that distinction entirely.

Quick answer: There are two completely different outcomes — extracting translated text (editable, copy-paste ready) versus getting back a translated image (layout preserved). Most free tools only do the first. For professional image assets like infographics or product images, you need a tool that does all three steps: OCR → translate → re-render.

Why "Translating an Image" Is a Three-Step Problem

Images don't contain text the way a Word document does. Text in an image has been flattened into pixels — there's no underlying text layer to select, copy, or translate directly.

To translate it, three things must happen in sequence:

1. OCR: The tool reads the pixel patterns and reconstructs the text as machine-readable characters
2. Translation: The extracted text is passed through a translation engine
3. Output: The result is either returned as plain text (step 2 output) or re-rendered back onto the image in the target language (step 3 output)

Most free tools stop at step 2. You get translated text, but the image formatting is gone. Tools that do step 3 — returning a translated image with the original layout — are rarer and generally paid.

What Type of Image Do You Have?

Screenshots — highest OCR accuracy (95%+). Standard fonts, perfect contrast. Any tool handles these well.

Photos of signs, menus, or printed documents — accuracy varies (70–90%) depending on lighting, angle, and distance. Flat, evenly lit, straight-on photos produce the best results.

Scanned documents — good accuracy (85–95%) when scanned at 300 DPI or higher . Below 200 DPI, character shapes start to blur.

Infographics with text over graphics — harder. OCR engines struggle when text sits on complex backgrounds or gradient fills.

Handwritten text — limited accuracy. Modern AI OCR reaches over 90% accuracy on clear handwriting but cursive remains challenging. Always verify manually.

Method 1: Extract Text Only, Then Translate Separately

Use this when you need editable translated text. You don't care about preserving the image.

Step 1: Extract text using OCR

Google Drive (free):

1. Upload your image to Google Drive
2. Right-click → Open with → Google Docs
3. Google Docs runs OCR and displays extracted text below the image
4. Copy the text

Google Lens (free, phone or Chrome):

1. On phone: open Google Lens, point at the image, tap text
2. On desktop: right-click any image in Chrome → "Search image with Google Lens" → select text
3. Copy the selected text

Step 2: Translate the extracted text

• Google Translate — free, 249 languages
• DeepL — better quality for European language pairs
• Doctranslate.io Text Translation — 100+ languages

Limitation: You get the translated words, but the image formatting is gone. If you need to preserve the visual layout, use Method 2.

Method 2: Translate the Image (Keep the Layout)

Use this when you need the image itself back with translated text — for publishing, sharing, or presenting.

Option A — Google Translate Image tab (free, basic)

1. Go to Google Translate → click the Images tab
2. Upload your image or paste a URL
3. Select source and target language
4. Google overlays translated text on the image
5. Click "Download translation" to save

What works: Free, instant, most major languages. What doesn't: Font mismatches, poor alignment on complex backgrounds. Not suitable for professional image assets.

Option B — Doctranslate.io Visual Translation (professional output)

1. Go to doctranslate.io/translation/visual
2. Upload your image (JPG, PNG, or screenshot)
3. Select source and target language from 100+ pairs
4. Download the translated image with the original layout preserved

What works: Handles complex backgrounds, mixed layouts, and images where text positioning matters — infographics, product images, app interfaces. Supports strong Asian language coverage (Vietnamese, Japanese, Chinese, Korean).

What doesn't: Requires a free trial (15 credits) or paid plan. Not the right tool for a one-off quick read.

Method 3: Image Inside a Document

If the image containing text is inside a PDF or scanned document — not a standalone image file — use Doctranslate.io Document Translation instead. It detects both native text and image-embedded text via OCR in the same pipeline.

Tool Comparison

Tips to Get Better Results

For screenshots: Use full resolution. PNG preserves sharpness better than JPEG for text-heavy images.

For phone photos: Shoot straight-on — even a 10–15 degree tilt reduces OCR accuracy. Ensure even lighting. Get close enough that individual letters are clearly distinguishable.

For scanned documents: Scan at 300 DPI minimum per archival standards. Most scanner apps default to 150–200 DPI.

Specify the source language if you can. Auto-detection works for common languages in single-language images. Mixed-language images benefit from manual selection.

Common Problems and Fixes

The translated text makes no sense. Almost always an OCR error, not a translation error. Improve image quality and re-run OCR before translating.

Only part of the text was translated. The OCR engine missed text with low contrast, decorative fonts, or text at an angle. Crop and zoom into the missed section and run OCR separately.

The translated image has text in the wrong position. The re-rendering step placed text without accounting for text expansion — translated text is often 20–40% longer than the source. Paid tools with layout-aware re-rendering handle this better.

Asian or Arabic text wasn't recognized. Not all OCR engines support non-Latin scripts equally. Google Lens and Doctranslate.io Visual Translation both have dedicated CJK and Arabic OCR support.

Use Cases by Audience

Researcher reading a foreign-language paper figure: Screenshot the figure → Google Lens to extract text → DeepL or Doctranslate.io Text Translation → paste into notes.

Marketer localizing an infographic: Upload to Doctranslate.io Visual Translation → get translated image back with layout preserved.

Traveler reading a menu or sign: Google Translate camera mode or Google Lens → instant read, no download needed.

Legal team reviewing a foreign document with embedded images: Upload the full document to Doctranslate.io Document Translation — OCR handles image-embedded text alongside native text.

Frequently Asked Questions

Can I translate text in an image for free? Yes. Google Translate's Images tab is free. Google Lens is free on Android, iOS, and Chrome. Both work well for quick reads. For layout-preserved output, a dedicated tool is needed.

What's the difference between image-to-text and image translation? Image-to-text (OCR) extracts characters and gives you plain text. Image translation goes further — it translates that text and either returns it as plain translated text, or re-renders it back onto the image in the correct position.

Can I translate handwriting in an image? Modern AI OCR handles clear printed handwriting with reasonable accuracy. Cursive, joined, or messy handwriting is significantly less reliable — expect errors and plan to review manually.

Does image quality affect translation accuracy? Directly and significantly. OCR accuracy drops sharply with low resolution, poor lighting, and skewed angles. A translation engine can only work with what OCR gives it — if OCR misreads characters, the translation will be wrong regardless of engine quality.

What to Do

For a quick read of a foreign-language image — sign, menu, screenshot — Google Translate's Images tab or Google Lens covers it at no cost.

For professional output — a translated infographic, a product image for a new market — test the specific tool on your actual image before committing to a plan.