PDF4QT 1.7.0.0 - OCR and Scanned Documents

PDF4QT 1.7.0.0 brings text recognition (OCR) and a complete workflow for scanned documents. Scanned pages can be prepared (crop, split, deskew, perspective correction), recognized, reviewed and corrected, and an invisible text layer is written into the document, so the scan becomes searchable and its text can be selected and copied. The scanned images can be compressed losslessly (JBIG2, CCITT G4) in the same step, which makes typical bitonal scans much smaller.

OCR is built on Tesseract. Fast models for English, Czech, Slovak, German, Spanish, Russian and Chinese are included, so recognition works offline out of the box; more than 100 further languages and the more accurate standard/best models can be downloaded from the application. Recognition is also available from the command line (PdfTool ocr, including batch processing), and the results can be exported as text, hOCR, ALTO or TSV.

The release also improves visible digital signatures, timestamps, the page content editor, annotation selection, measurements, redaction and text-to-speech, and fixes a number of rendering and form problems.

Highlights

• OCR of scanned documents with review, corrections and an invisible text layer (#85)
• Scan preparation: crop, split of double pages, deskew and perspective correction; compression of the scanned images (JBIG2 / CCITT G4)
• Bitonal scans are displayed without losing thin strokes, updated conversion to bitonal documents (#427, #425)
• Digital signatures with a visible signature and RFC 3161 timestamps (#417, #188)
• Selection of annotations - select, move, resize, rotate, copy and paste annotations (#252)
• Many fixes of the page content editor - inverted text, flipped or disappearing images, lost text position, corrupted vector graphics (#410, #211, #413, #238, #337, #203, #204, #196, #411)
• Insert page numbers, faster outline creation, improved measurements (#258, #422, #424)
• Redaction: choice of the color of the rectangles, inverted in dark mode (#409, #420, #419)
• Text-to-speech: no audio output fixed, locale add-ons and system voices (#397, #398)
• Faster text layout and rendering fixes (#430, #415, #393, #328, #299, #223, #431)
• Forms and checkboxes fixed (#392, #327)
• PDF4QT Diff fixes, including opening files on Linux (#278, #231)

The complete list of resolved issues is in RELEASES.txt.

Notes

• The Windows MSI installer is not code signed in this release (the code signing certificate has expired). The Windows packages and the AppImage have GitHub artifact attestations, they can be verified by gh attestation verify <file> --repo JakubMelka/PDF4QT.

Next version

The next version will be PDF4QT 2.0 with a new, professional design.