OCR stands for optical character recognition. At the most basic level, OCR converts a scanned document into a readable file by turning an image of text into actual data.
There’s more than just a simple conversion, though. There’s also the ability to make adjustments. With OCR, you can turn a static PDF or scanned image into something dynamic — like a searchable file, which can be used in various programs and shared as a downloadable file.
Scanning is the easiest way to preserve a document. But that’s only part of the story. Without OCR, your document can’t actually be searched for keywords. In addition, search engines won’t find your content automatically, and assistive technology won’t read it aloud (for people who are blind, for example). That means your scanned documents don’t contain machine-readable text unless they’re created with OCR.
This article explains what OCR does, when it’s needed, and how to apply it effectively using popular tools like Adobe Acrobat Pro, Foxit, and Acrobat Online.
What is OCR?
Optical character recognition (OCR) converts a scanned image of text into machine-readable text. The output isn’t a picture of the text; instead, it’s actual text that you can edit and use in different applications.
You’ll typically encounter OCR if you scan a paper document and then want to use the document within another program, such as Microsoft Word or Excel. You’ll also see OCR on images taken with a camera, image-only PDFs, and scanned documents, because these files don’t have any internal text.
OCR eliminates the need for manual data entry. Instead of re-entering information from a PDF or scanning it manually, OCR turns static content into something that you can adjust and share.
Without OCR, the text inside a document isn’t searchable. If you’ve ever tried to search for words in an image, you know that search engines can’t detect them. Searchable text is much easier to work with, too. For instance, you can extract specific portions of text and copy/paste them into other programs.
What OCR doesn’t do
In many cases, you might not even know that you’re working with OCR. Still, it’s important to understand what OCR can and can’t do.
OCR is great at turning a static document into something dynamic. But without OCR, a scanned document has no value beyond being an image.
It’s not possible to convert scanned documents to searchable formats without OCR. You can perform OCR processing on any scanned document, but it’s important to know how well the tool will recognize the text.
How OCR works
The OCR process involves several steps. Before characters are recognized, the layout of the page is analyzed to divide the document into sections that are appropriate for OCR. During the process, an OCR engine identifies individual characters, compares them to existing fonts, and tries to match the text to known patterns.
Before OCR takes place, there are some preliminary tasks that must occur, including:
Enhancing contrast between characters and the background
Removing noise and extraneous lines that may appear due to low-resolution scans
Correcting skew or rotation issues
Converting grayscale or color images to black and white
Binarization separates foreground objects, like letters, from background objects, like paper
As you might expect, the quality of the image directly affects the accuracy of OCR. As long as the original document was clear and legible, the chances of accurate OCR increase.
When is OCR inaccurate?
Some conditions and document qualities make OCR less accurate, sometimes to the point where a person would prefer to manually enter the text rather than use OCR. Here are some situations where OCR recognition may be difficult:
If you’re scanning low-quality documents, OCR becomes less accurate. While it’s technically possible to scan at 300 DPI, anything less than that increases the chance of errors.
OCR struggles to accurately identify handwriting. Unless you’re using a specialized application, like ScanSnap, it’s unlikely that OCR will identify handwriting with high accuracy.
Complex table structures may cause segmentation problems, leading to misinterpretation of data.
If your documents vary widely in structure, formatting, or language, standard OCR models might struggle to keep up. In those cases, you might require a customized OCR model.
Adobe Acrobat flags potential OCR “suspects.” These items include text that wasn’t recognized confidently enough for the system to label as OCR-generated. After processing, you’ll usually have an option to review and accept or reject these items.
Once processed, you should review both the text and its reading order to ensure accuracy.
Best practices for scanning
Many people assume that simply saving their documents as a PDF file means that they’re safe and secure. Unfortunately, the truth is that nothing about the original document ensures that it contains searchable, editable text. If you’d like to apply OCR before the scanning stage, here are our best practices:
Scan documents at 300 DPI to preserve quality.
Save your scanned files as a TIFF or PNG. These file types preserve quality while avoiding compression artifacts.
Don’t use JPEG for OCR input, since compression artifacts may interfere with OCR processing.
Applying OCR
Here’s how to apply OCR using the tools that we recommend, starting with Adobe Acrobat Pro.
Adobe Acrobat Pro
Open Adobe Acrobat Pro and click the All Tools icon. Next, choose Scan & OCR > Recognize Text: In This File. The program performs OCR processing and highlights any text that was either rejected or accepted as a suspect.
Adobe recommends that you review the OCR suspects after processing is complete. Any OCR-suspect items will be highlighted within the PDF document. Accept or reject each of these items based on your preference.
Foxit
Foxit recognizes text in scanned PDFs. Open the scanned PDF in Foxit, right-click on it, and choose Run OCR. Choose the destination folder for your new OCR-processed PDF and save the file.
Adobe’s online OCR tool
Adobe also includes OCR processing in its Acrobat Online tool, which enables you to upload your own files and download or share the resulting searchable PDF. To do so, visit Acrobat.com/ocr and select the Upload button. OCR is applied automatically, and you’ll be able to download or share the final version of your PDF.
One interesting side note: Recent versions of Acrobat convert raster-based PDFs into hybrid searchable PDFs. Hybrid PDFs combine text, images, and vector graphics in one file. They’re useful for PDFs that require additional processing, like extracting structured data.
Why OCR matters
OCR becomes essential in organizations or in high-volume document processing scenarios. If you have a high, repetitive volume of documents to process, such as hundreds of invoices or bank statements per month, then OCR helps speed things up and ensures consistency. Invoices, checks, bank statements, and loan agreements are common examples of documents that often appear in large numbers.
OCR makes sense if you have scanned or non-editable PDF documents. When performing auditing or credit analysis, for instance, consistency and speed become key requirements. OCR is the first step toward extracting structured data from these sources.
Manual entry doesn’t scale well. OCR provides an efficient alternative, although it still requires validation to ensure that the extracted data is correct.