Fix Incorrect Characters in Scanned PDF OCR Results
Question details
The user needs to fix an issue where certain characters in a scanned PDF are recognized as image objects instead of text during the OCR process.

- Product
- Device & OS
- not provided
- Scenario
- Performing OCR on a scanned PDF after modifying the document background.
- Observed behavior
- Specific characters like curly braces appear as image objects instead of text. Deleting the image removes the character, and OCR fails to correctly recognize the content.
Ensure you have the original, unedited version of the scanned PDF, as altering the background before applying OCR can significantly degrade text recognition accuracy.
Re-run OCR on the Original Unedited Scan
Applying OCR before altering the PDF background ensures maximum contrast and accuracy for text recognition.
When the background of a scanned document is edited before OCR, the contrast between the text and the background is often compromised. This causes the OCR engine to misinterpret complex characters, like curly braces, as arbitrary image objects.
Open the original, unedited scanned PDF file before any background changes or color adjustments were applied.
Access your PDF editor's OCR tool and ensure the correct document language and high-quality recognition settings are selected.
Execute the OCR process on the document to convert the scanned images into editable text.
Once the text is successfully recognized and verified, proceed to change or edit the PDF background as needed.
Enhance Scan Quality Before OCR
Improving the contrast and clarity of the scanned document can help the OCR engine properly identify complex characters.
Use WPS Office for Highly Accurate PDF OCR
WPS Office provides a powerful, built-in OCR feature designed to accurately recognize text from scanned PDFs and images, avoiding broken characters and unrecognized image objects.
- 1. Open your PDF: Launch WPS Office and open your scanned PDF document.
- 2. Access the OCR tool: Navigate to the 'Tools' tab on the top ribbon and select 'OCR' to initiate the text recognition tool.
- 3. Select output options: Choose 'Convert to Editable Document' or 'Extract Text', select the correct document language, and click 'Start'.

Frequently Asked Questions
Why do some characters turn into image objects during OCR?
This usually happens when the scan quality is low or the contrast between the text and background is poor. The OCR engine fails to identify the shape as a known text character and treats it as an embedded image instead.
Should I change the PDF background before or after running OCR?
You should always run OCR before changing the document background. Altering the background first can reduce text contrast, leading to poor recognition results and missing characters.
How can I fix OCR errors on just one page?
You can extract the specific page with errors, enhance its image quality or adjust the contrast separately, and run the OCR tool on that single page to achieve better accuracy before reinserting it into the main document.




