PDF 文字辨識 (OCR)
使用 OCR 光學字元辨識技術從掃描文件中擷取文字。
About OCR PDF
How OCR Extracts Text from Scanned PDFs
A scanned PDF is just images of pages. The text you see is pixels, not characters. OCR (Optical Character Recognition) analyzes those images and converts the visual text into actual, searchable, copyable text data.
OxygenPDF offers multiple OCR engines. Client-side engines like Tesseract and PaddleOCR run entirely in your browser using WebAssembly — your file never leaves your device. For more demanding tasks, cloud engines like Surya and GOT-OCR 2.0 send page images to a server for higher accuracy on complex layouts.
When You Need OCR
Any time you have a PDF where you can’t select or search the text, OCR is the fix.
Make Scans Searchable
Scanned contracts, receipts, and forms become searchable text. Find what you need without reading every page.
Extract Text for Reuse
Copy text from scanned documents into emails, spreadsheets, or other documents. No manual retyping.
20+ Languages Supported
English, Chinese, Japanese, Korean, Arabic, Russian, and more. Select the document language for best accuracy.
Smart Mode Detection
Smart mode checks for embedded text first. If the PDF already has selectable text, it extracts that directly without running OCR.
Multiple OCR Engines
Tesseract is the default — an open-source engine that runs in your browser via WebAssembly. Best for English documents and straightforward layouts. Fast, private, no download required.
PaddleOCR excels at Asian languages (Chinese, Japanese, Korean) and offers two variants: a fast model (~20 MB download) and a quality model (~165 MB) for higher accuracy. Florence-2 is an AI vision model that requires WebGPU (Chrome) and handles complex document layouts well.
For maximum accuracy on difficult documents — tables, math, mixed layouts — cloud engines like Surya (90+ languages) and GOT-OCR 2.0 send page images to a server. These trade privacy for accuracy.
How It Compares
OCR accuracy varies by engine and document type. The key differentiator is whether you get a choice between local and cloud processing.
| 功能特色 | OxygenPDF | Adobe Acrobat | Smallpdf |
|---|---|---|---|
| Client-side OCR (no upload) | Desktop only | ||
| Multiple engine options | |||
| 20+ languages | Limited | ||
| Smart mode (skip OCR if text exists) | |||
| Quality presets (fast/balanced/best) | |||
| Free to use | Limited |
Privacy-First by Default
Scanned documents often contain exactly the kind of content you don’t want on someone else’s server: tax forms, medical records, IDs, legal filings. Most online OCR tools upload your images for server-side processing.
OxygenPDF defaults to client-side engines that run in your browser. Your file never leaves your device. Cloud engines are available when you need higher accuracy, but they’re opt-in and clearly labeled.
PDF 文字辨識 (OCR): 常見問題
關於此工具及其運作方式的常見問題。
Pro
$9 單次付費