A scanned PDF is much harder to translate than a text-based PDF. A text-based PDF contains machine-readable text, whereas a scanned PDF contains images, which in turn contain text.
To translate a scanned PDF, you first need to accurately extract the text from those images. Only then can you translate it.
So, how do you translate a scanned PDF? First, let’s look at what makes a PDF “scanned.”
What is a scanned PDF?
Simply put, a scanned PDF contains images of a document rather than machine-readable text.
A text-based PDF is easy to handle. Even as a user, you can select the text, highlight it, copy it, search it, and edit it.
Scanned PDFs, on the other hand, are images created by scanning or photographing physical documents. You can’t select, copy, or edit the text. That’s why translators have a hard time with scanned PDFs.
So, how do you translate a PDF made of images rather than text? That’s where OCR comes into the picture.
What is optical character recognition (OCR)?
Optical Character Recognition (OCR) is a technology that detects text, characters, and elements in images and converts them into a machine-readable format.
To translate any scanned PDF, you first need OCR to detect the content and convert it into machine-readable text. Once you have the text, you can translate it into your target language.
A quality PDF translator not only provides accurate translations but also has high-quality OCR. The higher the OCR quality, the better it detects content in scanned pages. Poor OCR means inaccurate content detection, which in turn leads to translation errors.
Premium PDF translators also reconstruct the PDF formatting detected in the scanned pages, including tables, headers, and other visual elements.
How to translate a scanned PDF (step-by-step guide)
Here’s a detailed step-by-step guide to translating a scanned PDF:
Step 1: Pick a translation platform with built-in OCR
We have compared different PDF translation tools before. For highly accurate translation of business documents, premium tools like Smartcat work best.
Smartcat’s PDF translator comes with built-in OCR, supports over 280 languages, and preserves the document’s original formatting, so it handles scanned PDFs accurately.
Step 2: Upload the scanned PDF
Check the quality of the scanned PDF first. Clear scans make it easier for OCR to detect text accurately. For best results, scan at a resolution of at least 300 DPI.
Also make sure the scanned pages have clear, legible text, minimal visual noise, and good contrast between the text and background. Once you’re satisfied with the scan quality, upload the scanned PDF to the translator.
Step 3: Select the source and target languages
Select the original language of the scanned PDF and the target languages you want it translated into. You can select multiple target languages, and Smartcat translates the PDF into all of them.
Step 4: Click Translate
Next, hit the Translate button. Smartcat uses OCR to detect the text and formatting in the scanned pages, extracts them, and then translates the text using AI.
Step 5: Review the translation in Smartcat Editor
Once the AI translation is ready, review the translated content.
For scanned PDFs, you need two levels of quality checks:
OCR accuracy. Look for errors in OCR detection. OCR commonly mistakes certain characters, like 0 for O, 1 for I, or rn for m, so check for these basic extraction errors.
Translation accuracy. Check the translation of technical terms, brand names, and cultural nuances that the AI translation may have missed.
Step 6: Download your translated PDF file
Once you have checked and finalized the translation, download the PDF files for each target language.
How to improve OCR accuracy before translating a scanned PDF
Translation quality partly depends on the quality of the uploaded PDF. A clear, high-resolution scan gives better output than an illegible, low-resolution one.
Here are a few ways to get better OCR results:
1. Use high-resolution scans
Scan resolution significantly affects OCR accuracy and, therefore, translation quality. Low resolutions make text extraction difficult and lead to errors.
A good scan resolution starts at 300 DPI, so make sure your scanned PDF meets that. Repeated compression reduces resolution even further, so avoid it.
2. Straighten skewed pages
Neatly, horizontally aligned text improves OCR accuracy. Pages scanned at an angle make character and line detection difficult, because skew distorts the scanned image.
Avoid scanning pages at an angle. When photographing documents, keep the camera parallel to the page.
3. Improve contrast and lighting
For higher accuracy, OCR should be able to easily distinguish the PDF elements from the background.
A good scanned PDF means dark text on a clean, light background. Faded text, dark backgrounds, and discolored or patterned pages make for poor scans. The same applies to photographed documents: avoid shadows across the text.
4. Avoid blurry images and visual noise
Blurred characters and visual noise confuse OCR and make content extraction difficult. Stamps, signatures, handwritten annotations, folds, stains, and overlapping visuals all interfere with accurate recognition.
Common problems when translating scanned PDFs
Here are the most common problems you might encounter when translating a scanned PDF, and how to fix them.
Problem | Likely cause | What to do |
|---|---|---|
Missing text | OCR failed to recognize part of the page | Check the scan and rescan poor-quality sections |
Incorrect characters | Blurry or low-resolution scan | Use a clearer, higher-resolution source |
Wrong translation | OCR error or translation error | Compare the extracted source text with the scan first |
Broken tables | Complex document structure | Review the table and layout manually |
Untranslated image text | Translator doesn’t support OCR or image translation | Use an OCR-capable translation platform like Smartcat |
Incorrect reading order | Multi-column or complex layout | Verify extracted segments against the source |
Text overflow | Translation is longer than the source | Adjust the translated layout before publishing |
FAQs
Can Google Translate translate a scanned PDF?
Google states that text inside images and scanned PDF pages may appear in the output document but isn’t translated. It may translate some easily detectable text in images, but there’s no guarantee Google Translate will translate a scanned PDF.
Can ChatGPT translate a scanned PDF?
ChatGPT and other LLMs can work with scanned PDFs and translate them to a decent level. For a professional-grade PDF translation with OCR, high accuracy, and format preservation, a dedicated translator is a much better choice.
What resolution is best for OCR?
For reasonably good text detection in scanned pages, most platforms recommend a minimum resolution of 300 DPI.
Resolution is just one factor, though. Straight pages, good contrast, low visual noise, and clearly printed text also help produce better results.
Can OCR translate handwritten PDFs?
OCR doesn’t translate text itself; it recognizes text so that a translation system can process it.
Handwriting is generally harder for OCR to recognize than printed text, particularly when it is cursive or unclear. But if the handwriting is legible, OCR can detect and extract the text.
Is it safe to upload confidential scanned PDFs online?
Data protection depends on the translation platform you upload your scanned PDFs to. Enterprise-level platforms like Smartcat offer data encryption and other measures to safeguard confidential information in uploaded documents.
So, when uploading scanned PDFs with sensitive information, choose a reliable translation platform like Smartcat.
