Scanned pages

A page that is only a picture has its text read on the phone when you long press it.

A scan or a photographed document has no text of its own, only a picture of it. Its text is read on the phone, when you ask for it.

01 · How to copy from a scan
  1. Long press the text on the page.
  2. Reading the text… shows at the top for a second or two.
  3. The word under your finger is selected. From there, it works like any other page: see Copying text.
02 · What happens
  • Only the page you long press is read, and only then. Nothing runs when you open or scroll a PDF.
  • The reading is done by Tesseract, on the phone, for English and Romanian. The models are inside the app; no picture is sent anywhere.
  • The page is first evened out: made grey, its uneven lighting flattened, its contrast stretched. Then it is read two ways, and the reading that finds more text is kept.
  • A page read once stays read while the PDF is open.
03 · What to expect
  • Clean scans read almost letter for letter. A dark or blurred photo reads less: some characters can come out wrong.
  • Handwriting is not read.

Opensolr PDF Reader is open source and MIT licensed. Questions about the app belong on GitHub.

Opensolr PDF Reader Documentation