Scanned PDF OCR: Searchable Text and KrobClick’s Limits
A scanned PDF may look readable but refuse name searches or line copying because pages are stored as images. This guide helps choose OCR and verify numbers and Thai text against the original.

Image-only PDFs lack searchable character data
If you can scroll but cannot select letters with the mouse or find visibly present words with search, the PDF may contain only scanned images. Image text lacks searchable character data found in PDFs created directly from text documents. OCR adds selectable, searchable text.
W3C explains OCR for scanned PDFs as a way to create actual text from images of characters. Reading order, accuracy, and assistive-technology accessibility still need checking; visible text alone is insufficient.
First check whether it is a scanned image or a document that already contains text
- Select text: Open the PDF and try selecting a sentence. If words select and copy correctly, OCR may be unnecessary.
- Search for specific words: Search for a word in the middle of the page using the PDF reader. Finding it in place suggests an existing text layer, but accuracy still needs checking.
- Copy a sample table row: If the column order is reversed or numbers are missing, the PDF may contain actual text but a poor readable structure. Repeating OCR does not necessarily resolve the issue.
- See if a document has copying restrictions: Some files have permissions or security settings that restrict functionality. It should not always be interpreted as a scan file.
How OCR adds a text layer, and why the original image remains
In Adobe Acrobat, Recognize Text uses OCR to analyze the image and add a selectable, searchable text layer to the PDF while the scanned image remains visible. Check the recognized text against the original image.
Adobe explains the process for recognizing text in scanned documents and states that the system adds a searchable text layer. It also recommends keeping an original copy before editing.
Example steps in Adobe Acrobat
These steps come from developer documentation, not KrobClick hands-on tests. Menus and entitlements may differ by product or plan:
- Keep a copy of the original PDF to compare data after conversion
- Open the PDF in Acrobat and go to All tools → Scan & OCR
- Select In this file then specify the pages and document language using available options
- Check the settings, then press Recognize Text and save a copy of the output
- Find the words you read and copy the sample text to make sure the text layer matches the image.
Adobe also provides Enhance Scanned Documents guidance for improving scanned documents. Check the OCR result against the original image after processing.
Check supported languages for that version, especially Thai. Do not assume Thai always appears. If the document language is unavailable, use a supporting tool and test a nonsensitive copy first.
Hypothetical example: OCR misreads just one digit and changes its meaning.
Suppose a scanned receipt actually says ยอดรวม 1,080.00 บาท but OCR reads it as ยอดรวม 1,030.00 บาท . The 50-baht difference is a recognition error, not a change to the original. Copying it into a spreadsheet without checking transfers the wrong amount. The document and figures are hypothetical examples, not test results from any tool.
| What to check | Errors that may be encountered | How to check |
|---|---|---|
| Name and document number | The letter O and the number 0 or I and 1 are interchanged. | Compare each location against the original image |
| Amounts and dates | Wrong number, decimal point, or minus sign | Check every important item and total. Then randomly check other additional items. |
| Thai language | Missing vowels or tone marks, or incorrect word segmentation | Choose a multiparagraph sample and check spelling |
| Multi-column table | Reads across columns or reverses row order | Copy it as text and read it in the actual order. |
Searchable does not equate to accessible for everyone.
Searchable PDFs may still lack correct headings, table structure, and reading order, particularly for screen readers. W3C recommends checking text completeness and order after OCR and creating or fixing necessary accessibility tags in supported tools.
If the original word-processing file is available, create the PDF from actual text with properly structured headings and tables, especially for documents distributed widely. This avoids converting text to images and then using OCR to recover it.
Review privacy before uploading scanned documents
Scanned PDFs may contain identification numbers, contact information, statements, or contracts. Uploading them to an external website for processing may send information to the provider's systems. Check the privacy policy, file storage, and processing location, and request permission from the data owner when necessary. Local processing may reduce the need to send files elsewhere, but consider the software's permissions and security too.
KrobClick’s image/PDF converters do not create an OCR text layer. The guide Reduce the PDF size to a specified limit addresses file size and readability rather than adding searchable text.
In KrobClick PDF Editor, open a scanned PDF, choose Scan & OCR, and recognize Thai or English text. The editor processes the file in your browser and skips pages that already contain text. Download the result, then search and copy text to check it against the original image.
If a page contains both selectable text and a scanned image, this editor skips the entire page when it finds existing text, even if the image text is not searchable. If there is no result to download, keep the original and try another tool that supports mixed pages. If there is a result, download it and try selecting and searching for text in the image on each page. If it is still missing, do not assume OCR is complete. Keep the original, try another suitable tool, and check the result again.
For documents containing Thai, choose Thai and English because English mode omits Thai from the text layer. OCR can make mistakes, and signed PDFs or files with editing restrictions may not work with this feature. The program and OCR models download from the site before processing in your browser. See details about saved work in the Privacy Policy.
Good OCR must be verifiable, not merely copyable
Identify the PDF type, use OCR only when necessary, back up originals, evaluate important languages and numbers, and check accessibility and privacy before sharing. Faster searches are no substitute for checking that the text matches the actual image, especially for financial or other important documents.
Sources
- Adobe Acrobat: Recognize text in scanned documents
- W3C WAI: PDF7 Performing OCR on a scanned PDF document
- Adobe Acrobat: Enhance scanned documents
Methods refer to Adobe and W3C documentation reviewed on October 13, 2026. OCR was not tested on real user files. Capabilities, languages, and menus may vary by version or license. Check data transmission and privacy before using external services.