Private local OCR workspace

OCR Recognition

Recognition accuracyText detectionPage segmentationWorking-image rotation

Resolution presets change the actual OCR rendering DPI. They do not guarantee accuracy. Any existing selectable text skips a page by default; use Force OCR for scans with a text footer. Existing text remains in the output, so forced recognition can add duplicates.

Working Image Preprocessing

OCR rendering resolution 300 DPI

Large pages are downscaled to the device pixel budget. The actual DPI is included in each page report. These settings affect recognition images only; original PDF images are not recompressed or replaced.

Grayscale PreprocessingRemove color from the working image before recognition.Auto Deskew / Auto Orientation — UnavailableAutomatic angle correction is not enabled for searchable output because coordinate mapping is not yet verified. Use the working-image rotation above for sideways text.

Contrast 1.00×Noise reductionImage binarization

Smoothing can remove fine text detail. Start with Balanced and change filters only when needed.

Text & Layout

Original appearanceExisting text retainedInvisible OCR text layer for PDF output

Preserve Line Breaks in TXTKeep the line breaks returned by the recognition engine.Normalize Whitespace in TXTCollapse repeated spaces and trim line edges. Recognized words are not corrected or invented.Editable Table Reconstruction / Automatic Text Correction — UnavailableAutomatic segmentation can recognize multiple text columns. This tool does not promise editable tables or infer missing words.

Document Preservation

Preserve Supported HyperlinksInternal links need retained destinations; supported web links are copied.Preserve Standard AnnotationsCommon non-widget annotations are copied. Document actions and JavaScript are omitted.Rebuild Supported BookmarksRetain supported titles and hierarchy using Fit to Page destinations.Preserve Basic MetadataCompatible source title, author, subject, and keywords.

PDF output uses the shared page-copy engine. Detected forms require flattening consent. Existing signature validity, accessibility tags, attachments, and XMP cannot be preserved; detected features require confirmation. TXT-only processing leaves the original unchanged.

Output & Reports

Output formatCustom filename

Filename: document_ocr.pdf

Include OCR Confidence ReportOffer a local JSON report with actual confidence values. Confidence is not a guarantee of text accuracy.Include Per-page OCR SummaryInclude each selected page’s source, status, characters, actual DPI, and review flags in the report.

Performance

Memory profileConcurrent OCR pages

One reusable recognition worker processes one page at a time. High resolutions need more memory. No document is stored in a persistent browser cache.

Give your scanned pages a voice.

How would you like to use your scan?

Choose your pages

Original PDF

Recognized Text