Forum Discussion
Azure Form Recognizer Redaction Issue with Scanned PDFs and Page Size Variations
Hi all, I’m working on a PDF redaction process using Azure Form Recognizer and Azure Functions. The flow works well in most cases — I extract the text and bounding box coordinates and apply redact...
How about this:
- Normalize DPI Before Processing – If the scanned PDFs have inconsistent DPI settings, try standardizing them using an image processing tool before passing them to Form Recognizer.
- Adjust Bounding Box Scaling – Some users have resolved similar issues by dynamically adjusting bounding box coordinates based on the detected page size.
- Use a Fixed Template Approach – If possible, define a standard template for expected document sizes and adjust the redaction logic accordingly.
- Check for OCR Misalignment – OCR engines sometimes misinterpret scanned text positioning. Running a preprocessing step to enhance text clarity might help.
- Explore Azure AI Document Intelligence – Azure Form Recognizer was renamed to Azure AI Document Intelligence, which may offer improved handling for scanned documents.
Document Intellegence scanned PDF or PNG to editable pdf - Microsoft Q&A