Hidden and Off-Page Text Removal and Page Reconstruction

Applies to: Pro and Enterprise for the removal option; analysis is available in all editions

What PDFSanitize detects

During analysis, PDFSanitize examines extracted words and records two specific conditions:

  • Text whose letters use an invisible text-rendering mode;
  • Text whose bounding box is completely outside the visible page boundary, with a small positional tolerance.

Findings identify affected pages and can record the detected regions.

Preserve Quality behavior

When Remove hidden / off-page text is enabled in Preserve Quality mode, PDFSanitize determines which pages contain one of the detected conditions. Only those pages are reconstructed from rendered content.

Before processing, the application reports how many pages will be reconstructed and asks for confirmation. Unaffected pages retain their normal PDF text and vector content.

Reconstruction details

Reconstructed pages are rendered at 300 DPI, placed back at the original page dimensions, and saved without the original page content streams, resources, or annotations. The final document is then structurally sanitized again.

Whether annotations or form appearances are included in the rendering depends on the active policy. The predefined Strict Privacy and Maximum Assurance profiles remove annotations and forms, so those interactive structures are not preserved as normal source objects.

Maximum Assurance behavior

Maximum Assurance reconstructs every page regardless of whether hidden text was detected. This is the preferred predefined profile when the goal is to remove dependence on original page-level structures rather than only the detected hidden-text regions.

Detection limitations

Hidden-information techniques are not limited to invisible text or off-page text. PDFSanitize analysis does not represent a proof that every form of concealed data, steganography, image-layer information, semantic disclosure, or reader-specific behavior has been found.

Use the detection results as one input to the disclosure decision, and visually review high-risk outputs after reconstruction.