How to make a PDF searchable: OCR for scanned PDFs

Scanning paper documents into PDFs removes the need for filing cabinets, but a scan is a picture of a page. A PDF that only contains that picture has no text for software to work with:

• Searching for a word returns no results.
• Text can’t be selected, highlighted, or copied.
• Redacting a term means finding and covering every instance by hand.
• Search tools and document management systems can’t index the content.

Making the PDF searchable fixes this. The process is optical character recognition (OCR).

What is a searchable PDF?

A searchable PDF contains two layers on each scanned page: the original image, and a text layer with the recognized words. The text layer is invisible, and each word is positioned over the matching word in the image. A reader displays the image, while search, selection, and copy operations use the text underneath.

Because the image stays in place, the page looks exactly as it did before OCR. That’s the difference between a searchable PDF and a converted text document: OCR adds text to the PDF without changing its appearance.

PDFs created digitally — exported from a word processor or generated by an application — already contain real text and don’t need OCR.

How to check whether a PDF is searchable

Open the PDF and try one of these checks:

1. Search for a word that appears on the page. If the viewer finds no matches, the page has no text layer.
2. Try to select a line of text. If the selection draws a box over the whole page instead of highlighting words, the page is an image.
3. Copy text from the page and paste it elsewhere. Missing or garbled text points to a page without a usable text layer.

A document can mix both kinds of pages, such as a digital report with a scanned signature page appended. Check each kind of page, not only the first one.

How OCR makes a PDF searchable

OCR software works through three stages to turn a scanned page into a searchable one.

Find the text on the page

The software first identifies where text exists in the image. Image processing such as binarization(opens in a new tab), skew correction, and noise reduction raises the contrast between text and background, which makes text regions easier to detect.

Recognize the characters

Each line of text is then analyzed to determine its characters. Fonts, sizes, and handwriting vary widely, so modern OCR engines use machine learning(opens in a new tab) trained on large sets of examples instead of fixed rules. The language setting matters here: An engine configured for the document’s language recognizes its characters and words more accurately.

Embed the text in the PDF

Finally, the recognized text is written into the page content as invisible text. Each word is scaled and positioned to sit over its image in the scan. The PDF format supports text that’s in the page content but not drawn, so readers can search and select it while displaying the original image.

How to make a PDF searchable with Nutrient

Nutrient provides OCR in several products. The right one depends on where the PDFs are processed.

Use the OCR API

The Nutrient OCR API makes PDFs and images searchable through a single HTTP request, with no OCR engine to install. Send the file to the /build endpoint with an ocr action and the document’s language:

curl -X POST https://api.nutrient.io/build \
  -H "Authorization: Bearer <YOUR-API-KEY>" \
  -o result.pdf \
  --fail \
  -F scanned=@document.pdf \
  -F instructions='{
      "parts": [
        {
          "file": "scanned"
        }
      ],
      "actions": [
        {
          "type": "ocr",
          "language": "english"
        }
      ]
    }'

The returned result.pdf has searchable, selectable text. The scanned PDF OCR tutorial covers account setup and supported languages. The OCR API guide for JavaScript, Python, PHP, and Java shows the same request in other languages.

Use the .NET SDK

For applications that process files on their own servers, Nutrient .NET SDK runs OCR locally. This example makes every page of an image-based PDF searchable in one call:

using GdPicturePDF pdf = new GdPicturePDF();
pdf.LoadFromFile(@"input_image_based.pdf");
pdf.OcrPages("*", 0, "eng", "", "", 200);
pdf.SaveToFile(@"output.pdf");

The PDF-to-searchable-PDF guide also shows page-by-page processing and the OCR parameters. To start from scanned images instead of a PDF, see converting images to searchable PDFs.

Use Document Engine or the mobile SDKs

Document Engine applies the same ocr action on the server, either when a document is uploaded or to documents it already stores. In mobile apps, the iOS and Android SDKs add OCR through an OCR library and language packs bundled with the app. A scanned document becomes searchable without leaving it. On iOS, OCR runs entirely on the device.

Get accurate OCR results

OCR quality depends mostly on the input. The OCR API tutorial recommends:

• Resolution — Scan at 200–300 dots per inch (DPI). Higher resolutions don’t necessarily improve accuracy.
• Clarity — Use clean, high-contrast scans without heavy compression, noise, or skew.
• Language — Set the document’s language, and specify every language a multilingual document uses.

After OCR, spot-check the result by searching for a few words from different parts of the document. Names, numbers, and table contents are the most likely to contain recognition errors.

What a searchable PDF enables

With a text layer in place, the scanned document supports the same text operations as a digital one:

• Search — Search within the document, or index archives so documents can be found by their content.
• Annotation — Highlight, underline, and mark up text with annotations.
• Redaction — Search for a term and redact every match instead of covering each instance by hand.
• Copy and extraction — Copy passages or extract the text for other systems.

A searchable PDF isn’t automatically an accessible PDF. Screen readers also need tags that describe headings, lists, tables, and reading order. PDF/Universal Accessibility (PDF/UA) covers those requirements, and OCR is usually the first step toward meeting them for scanned documents.

FAQ