PDF Scan Enhancer — Clean Every Page Automatically
Clean up a photo-scanned PDF — whiten background, darken text, sharpen edges.
Drop your scanned PDF here
Every page is cleaned automatically — white background, dark text, sharp edges
Choose PDFLoading preview…
Higher = more aggressive background whitening
About PDF Scan Enhancer
Upload a PDF made from phone photos or a poorly-calibrated scanner and get back a clean, readable version. The tool processes every page automatically — converting to grayscale, whitening the background using histogram auto-levels, boosting contrast, and optionally sharpening edges. Download the enhanced PDF in one click.
Use Cases
Cleaning up PDFs made from phone photos of books or answer sheets, fixing poorly-scanned documents with grey backgrounds, enhancing old scanned archives, preparing scanned PDFs for printing or submitting to portals that require clear text.
How It Works
Uses PDF.js to render each PDF page to a canvas at 144 DPI. Applies grayscale conversion, histogram auto-levels (background whitening), contrast boost, and optional unsharp-mask sharpening. Embeds the processed JPEG images into a new PDF using pdf-lib and triggers a browser download.
FAQs
Will this work on a PDF made from phone photos?
Yes — that is the primary use case. The auto-levels step removes uneven lighting and shadows from phone photos, producing clean white backgrounds and dark text.
Does the output PDF stay editable?
No. Each page is converted to a JPEG image and embedded in the output PDF. The text is no longer selectable. This is the same result you get from a real scanner that does not run OCR.
Will the output file be larger or smaller than the original?
It depends on the source. Phone-photo PDFs that embed high-resolution photos will produce smaller output (JPEG at 144 DPI is efficient). PDFs that already have text layers may produce larger output since text is replaced with rasterised images.
How long does it take for a 25-page PDF?
Approximately 20–40 seconds depending on your device, as each page is rendered, processed, and re-embedded in sequence. A progress bar shows the current page.