OCR 依赖 Tesseract，需另行准备；Linux 端需系统已装 libsane 及 GTK 3.20+；README 未提供命令行安装方式，需从官网下载安装包
README 未给出安装命令，仅指向 https://www.naps2.com/download 下载 硬件要求按 README 说明
README 未说明显存或 GPU 要求；系统要求为 Windows 7+（x64/x86）、macOS 10.15+（x64/arm64）、Linux（x64/arm64，需 GTK 3.20+、glibc 2.27+、libsane）。
跨平台文档扫描应用，把纸质文件扫成 PDF 等格式，操作尽量简单
跨平台文档扫描工具，支持 TWAIN、WIA、eSCL 和 SANE 协议，把纸质文件扫成 PDF 并内置 OCR 识别，Windows、Linux、macOS 都能用。
核心能力
- 支持 WIA、TWAIN、SANE、ESCL 四类扫描仪接口
- 扫描后可按需整理页面顺序再导出
- 导出为 PDF、TIFF、JPEG、PNG 格式
- 借助 Tesseract 对扫描件做 OCR 文字识别
- 提供 NAPS2.Sdk 供开发者在自己的程序里集成扫描
容易踩的坑
内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README
- v8.6.08.6.0最新正式版2026-10-25 · 2026-10-25 21:28 UTC
Changes in 8.6.0:
- New combined Settings window
- Added toolbar customization (right-click the toolbar > "Customize Toolbar")
- Batch Scan/Auto Save: Show an error if the file path is missing an extension
- Added a "Retry" button for errors when saving
- "Edit with" now shows the icon of the selected app and is visible in the Preview window
- Escl: "Manual IP" now shows the connection status
- Mac: Native "Liquid Glass" sidebar in the main window
- Mac: Added "Apple Driver Implementation" to advanced profile settings (File or Memory transfer)
- If scanned pages aren't showing, you can try switching to "File Transfer"
- Improved quality of some icons with hi-dpi screens
- Added placeholder support (e.g. "$(YYYY)") to PDF metadata
- Sdk: ImageSharp 4.x compat
- Bug fixes:
- Auto-crop: Better handling for black backgrounds
- Auto-crop: Remove padding added by scanner drivers when the scan size exceeds the sensor
- Compress: The preview now shows the effects of JPEG quality
- Windows: Potential fix for crashes in the save dialog
- Windows: NAPS2.Portable.exe now passes command-line arguments (e.g. for opening a file)
- Windows: Fixed flicker on startup
- Mac: Fixed scanning when run from Shortcuts (or with a long TMPDIR)
- Linux: Fixed "Feeder" scanning both sides with some scanners (e.g. Kodak kds_s2000 driver)
- Linux: Fixed layout issues
- Sane: Fixed feeder/duplex selection with some scanners
- Wia: Fixed an error with devices that don't report their scan area size
- Fixed file auto-numbering behavior in some cases
- Fixed an error when using Scale with "Stretch/Crop to page size"
- Fixed an error with "Always Ask" when selecting a different driver type
- Fixed an issue with the "All Images" file filter and .jpg files
- Fixed some PDF import crashes
- Fixed a broken link to icons in the About window
- OCR: Fixed an error on some Linux OS's
- Sdk: Fixed TWAIN hangs with NAPS2.Sdk.Worker.Win32 by running the worker on an STA thread
- Sdk: Fixed spare worker processes not stopping when disposing ScanningContext
- v8.5.18.5.12026-10-12 · 2026-10-12 18:20 UTC
查看更新说明
Changes in 8.5.1:
- Windows: Fixed an issue with Smart App Control blocking naps2_jpeg.dll
- v8.5.08.5.02026-10-12 · 2026-10-12 03:21 UTC
查看更新说明
Changes in 8.5.0:
- Added "Auto-Crop" feature (for scanning varying-size items like receipts, photos, etc.)
- Click the button in the Crop window
- Or check "Auto-crop scanned pages" in your Advanced profile settings
- Significant PDF compression improvements
- JPEG images are now encoded with Jpegli (~25% smaller)
- Black and white images now use JBIG2 compression (~30% smaller)
- OCR adds less size (~80% less overhead)
- Fixed an issue where edited images are saved at the wrong quality level
- New "Compress" window (under the Image menu)
- Allows changing an image's resolution, dimensions, color depth, and quality
- Better OCR text alignment (now more like Tesseract)
- Added "Fix white balance and remove noise" to advanced profile settings
- This is equivalent to "Document Correction" under the Image menu
- Also made some improvements and speedups to Document Correction
- New profiles now default to the page size for your locale (Letter or A4)
- This can be overridden in appsettings.xml (the base value is now SystemDefault instead of Letter)
- Mac: Fixed an issue with the SANE driver
- Escl: Fixed resolution being lost from some devices
- Escl: Fixed some devices (e.g. Kyocera) staying busy after a scan
- Sdk: Fixed a duplicate WIA page start event
- Added "Auto-Crop" feature (for scanning varying-size items like receipts, photos, etc.)
- v8.4.18.4.12026-10-03 · 2026-10-03 21:00 UTC
查看更新说明
Changes in 8.4.1:
- Linux: Fixed a crash when "Show Profiles Toolbar" is enabled
- Mac: Fixed a scanning issue with Samsung scanners and Apple Driver
- Escl: Fixed an issue connecting to the wrong IP with mDNS responders
- v8.4.08.4.02026-10-01 · 2026-10-01 22:10 UTC
查看更新说明
Changes in 8.4.0:
- Added "Manual Duplex" (under "Reorder") as a clearer way to combine front and back sides scanned separately
- This is meant to replace the "Interleave" buttons, which will be removed in a future version
- Checking "Reverse back sides" is equivalent to using "Alternative Interleave"
- Unchecking "Always show preview" will add a separate button for one-click actions like before
- Admins can use "HideManualDuplex"/"HideInterleave" to control visibility
- Crop: Added a middle handle for moving the whole selection
- Crop/Split/Combine: Added arrow key controls
- Added help text to some image forms
- Linux: Added a "Theme" setting to manually switch between Light and Dark mode
- Mac: Custom page sizes are no longer shown for feeders with the Apple driver (as they don't work)
- Mac: Fixed keyboard shortcuts for non-QWERTY keyboard layouts
- Mac: Fixed flatbed/feeder detection with the Apple driver
- Fixed the "Maximum quality" option not being retained after cropping
- Fixed the "Black and White" keyboard shortcut not being editable in Keyboard Shortcuts
- Fixed an issue with email attachments not working when Thunderbird is already open
- Console: The --device option can now match the device ID as well as the name
- Sane: KeyValueOptions set in profile XML now support true/false and are retained when editing a profile
- Escl: Fixed a misspelled feeder status value and added an admin page for better compatibility
- Sdk: Added MetadataCaps.ProtocolVersion
- Added "Manual Duplex" (under "Reorder") as a clearer way to combine front and back sides scanned separately
用户点评
登录用户的真实使用记录，可按版本和硬件筛选共 0 条
还没有人点评。第一个说出真实体验的人最有价值。
NAPS2 - Not Another PDF Scanner
NAPS2 is a document scanning application with a focus on simplicity and ease of use. Scan your documents from WIA, TWAIN, SANE, and ESCL scanners, organize the pages as you like, and save them as PDF, TIFF, JPEG, or PNG. Optical character recognition (OCR) is available using Tesseract.
System requirements:
- Windows 7+ (x64, x86)
- macOS 10.15+ (x64, arm64)
- Linux (x64, arm64) (GTK 3.20+, glibc 2.27+, libsane)
Visit the NAPS2 home page at www.naps2.com.
Other links:
NAPS2.Sdk (for developers)
NAPS2.Sdk is a fully-featured scanning library, supporting WIA, TWAIN, SANE, and ESCL scanners on Windows, Mac, and Linux. Read more.
Build Instructions
Looking to contribute to NAPS2 or NAPS2.Sdk? Have a look at the Github wiki for build instructions and more.
License
NAPS2 is licensed under the GNU GPL 2.0 (or later). Some projects have additional license options:
- NAPS2.Escl.* - GNU LGPL 2.1 (or later)
- NAPS2.Images.* - GNU LGPL 2.1 (or later)
- NAPS2.Internals - GNU LGPL 2.1 (or later)
- NAPS2.Sdk - GNU LGPL 2.1 (or later)
- NAPS2.Sdk.Samples - MIT
PaddleOCR
PaddlePaddle·Python
轻量级 OCR 工具包，能把图片和 PDF 转成结构化数据，支持 100 多种语言，方便接入 LLM 和 RAG 流程。
OCR+4文档解析 / OCR
92516+13012kv3.7.02026-09-16 03:31 UTC
MinerU
opendatalab·Python
把 PDF、Office 等复杂文档转成 LLM 可直接用的 Markdown/JSON，内置版面分析和 OCR，专为 RAG 与 Agent 工作流准备语料。
文档解析+4文档解析 / OCR
83425+1706.9kmineru-4.1.7-released2026-11-10 11:15 UTC
tesseract
tesseract-ocr·C++
老牌开源 OCR 引擎，用 LSTM 识别图片中的文字，支持上百种语言，适合做文档扫描和文本提取。
OCR引擎+4文档解析 / OCR
77437+3811k5.5.42026-11-09 11:40 UTC
docling
docling-project·Python
把 PDF、DOCX、HTML 等文档解析成结构化 Markdown 或 JSON，方便喂给大模型和 RAG 流程。
文档解析+4文档解析 / OCR
70364+1245.3kv2.149.02026-11-10 07:57 UTC