What Is Document Digitization?什么是文档数字化?
Place this specific workflow in context with the unstructured data processing and document intelligence guide, which connects the definitions, alternatives, validation steps, and related implementation guides.
可通过非结构化数据处理与文档智能指南理解本专题在整体流程中的位置;该指南串联了定义、替代方案、验证步骤与相关实施文章。
Document digitization is the controlled conversion of physical records into digital representations that remain complete, usable, identifiable and traceable to their source. A defensible program includes selection, preparation, capture, quality assurance, quality control, optional OCR, metadata, packaging, integrity checks, delivery and preservation.
文档数字化是把实物记录受控转换为数字再现物,使其保持完整、可用、可识别并能追溯到来源。一套可辩护的项目包括选择、准备、捕获、质量保证、质量控制、可选的OCR、元数据、打包、完整性校验、交付与保存。
The digital image is evidence of a capture event, not proof that every page, mark or relationship survived correctly. A usable result must preserve the information required by the intended use, keep page and item relationships, carry enough provenance to explain how it was made, and pass documented acceptance checks. The original may still need to be retained under law, policy, contract or historical-value decisions.
数字图像只是一次捕获事件的证据,并不能自动证明每一页、每个标记或相互关系都被正确保留。可用结果必须保存预定用途所需的信息,保持页与项目的关系,携带足够的来源信息来解释其生成方式,并通过有记录的验收检查。原件仍可能因法律、政策、合同或历史价值判断而需要保留。
Quick answer: begin with purpose, material and acceptance criteria. Inventory and prepare the originals; test representative samples; choose capture and format requirements by use; capture with controlled equipment; reconcile items and pages; inspect image, OCR and metadata quality; generate a manifest and checksums; then preserve the approved master package while delivering fit-for-purpose access copies.
快速回答:先确定目的、材料和验收标准。盘点并准备原件,用代表性样本测试,按用途选择捕获与格式要求;使用受控设备捕获,对账项目和页数,检查图像、OCR与元数据质量,生成清单和校验和;最后保存获批主文件包,同时交付适合使用的访问副本。
Document Digitization Is More Than Scanning or OCR文档数字化不只是扫描或OCR
Search results often blur scanning, digitization, OCR, data extraction and digital transformation. They solve different layers of the problem. Treating them as interchangeable produces image-only archives, missing pages, untraceable derivatives and premature automation.
搜索结果经常混用扫描、数字化、OCR、数据提取和数字化转型,但它们解决的是不同层面的问题。把这些概念互换使用,容易形成只有图像的档案、漏页、不可追溯的副本和过早自动化。
| Term术语 | Primary output主要输出 | What it does not prove不能证明什么 |
|---|---|---|
| Scanning / image capture扫描/图像捕获 | Raster image of a page or object页面或实物的栅格图像 | Completeness, correct order, readable text, metadata or preservation完整性、正确顺序、文本可读、元数据或长期保存 |
| OCROCR | Machine-readable text linked to an image与图像关联的机器可读文本 | Text accuracy, layout, reading order or accessibility文本准确性、版面、阅读顺序或可访问性 |
| Data extraction数据提取 | Selected fields or entities选定字段或实体 | Faithful preservation of the full record对完整记录的忠实保存 |
| Document digitization文档数字化 | Controlled master, derivatives, metadata and integrity evidence受控主文件、副本、元数据与完整性证据 | Automatic legal equivalence or permission to destroy originals自动取得法律等效性或销毁原件的许可 |
| Digital transformation数字化转型 | Changed service, process or operating model改变后的服务、流程或运营模式 | That underlying records were captured well底层记录已被高质量捕获 |
Choose Capture Requirements by Material and Intended Use按材料与预定用途选择捕获要求
A number such as pixels per inch is not a complete specification. Define the smallest significant detail, reproduction scale, color or grayscale need, bit depth, tonal response, sharpening, compression, acceptable cropping, geometric distortion and handling constraints. Calibrate and monitor the full imaging system—device, lens, lighting, target, software, display and operator—not merely the menu setting.
每英寸像素数之类的数字并不是完整规格。应定义最小重要细节、再现比例、彩色或灰度需求、位深、阶调响应、锐化、压缩、可接受裁切、几何失真与操作限制。需要校准和监测整个成像系统,包括设备、镜头、照明、标靶、软件、显示器和操作员,而不仅是菜单设置。
| Question问题 | Why it changes the specification为何会改变规格 | Test evidence测试证据 |
|---|---|---|
| What detail must remain?必须保留什么细节? | Fine handwriting, halftones, stamps and marginal marks need different resolving performance.细小手写、网点、印章和边注需要不同的解析性能。 | Representative target and material samples at normal viewing scale代表性标靶与材料样本,并按正常观看比例检查 |
| Does color carry meaning?色彩是否承载意义? | Annotations, fading and paper tone may disappear in bitonal capture.批注、褪色与纸张色调可能在二值捕获中消失。 | Color target, neutral balance and visual comparison色彩标靶、中性平衡与视觉比较 |
| Can the item contact glass?原件能否接触玻璃? | Bindings, fragile media and surface texture may require overhead or specialist capture.装订、脆弱介质与表面纹理可能需要俯拍或专业捕获。 | Condition assessment and approved handling method状况评估与获批操作方法 |
| Will the image support OCR?图像是否用于OCR? | Skew, blur, bleed-through, noise and compression can reduce recognition quality.歪斜、模糊、透印、噪声和压缩会降低识别质量。 | OCR trial on difficult language, layouts and conditions对困难语言、版面和状况进行OCR试验 |
Use a material-specific, measurable and monitored specification rather than claiming quality from a nominal scanner setting. NARA's Digitization Quality Management Guide connects requirements, objective testing, inspection and acceptance within a quality-managed program. Apply the institutional or regulatory standard that governs the collection.
应使用按材料制定、可测量且受监测的规格,而不能根据扫描仪标称设置声称质量。NARA的数字化质量管理指南把要求、客观测试、检查与验收纳入质量管理项目。请应用实际管辖该集合的机构或监管标准。
Run a Repeatable Document Digitization Process执行可重复的文档数字化流程
- 1. Charter and sample.1. 立项与抽样。 Approve scope, rights, retention, uses, owners, risks and acceptance criteria; test every representative material family.批准范围、权利、保留、用途、责任、风险与验收标准,并测试每个代表性材料家族。
- 2. Register the batch.2. 登记批次。 Freeze the expected inventory, hierarchy, identifiers, sides and page order before capture.捕获前冻结预期清单、层级、标识、正反面和页序。
- 3. Prepare and track.3. 准备与跟踪。 Assess condition, perform authorized reversible preparation and preserve item identity through every workstation.评估状况,执行获批且可逆的准备,并在每个工作站保持项目身份。
- 4. Calibrate and capture.4. 校准与捕获。 Verify device state and targets, capture all required surfaces, and record operator, device, date and settings.验证设备状态与标靶,捕获所有要求的表面,并记录操作员、设备、日期与设置。
- 5. Perform immediate QC.5. 立即质检。 Detect missing, duplicate, inverted, blurred, clipped, skewed or contaminated images while rescanning remains practical.在仍便于重扫时发现遗漏、重复、倒置、模糊、裁切、歪斜或污染图像。
- 6. Create derivatives.6. 创建副本。 Generate OCR, searchable packages, thumbnails or access files from the approved image source with versioned transforms.从获批图像源使用版本化转换生成OCR、可检索包、缩略图或访问文件。
- 7. Describe and package.7. 描述与打包。 Attach descriptive, structural, administrative, rights, preservation and technical metadata; create a manifest and checksums.附加描述、结构、管理、权利、保存与技术元数据,并创建清单和校验和。
- 8. Validate, accept and preserve.8. 验证、验收与保存。 Reconcile counts, validate files and relationships, resolve defects, sign the acceptance record, ingest masters and monitor integrity.对账数量,验证文件与关系,解决缺陷,签署验收记录,接收主文件并监测完整性。
Build Quality In, Then Inspect the Output先把质量纳入流程,再检查输出
Quality assurance designs the process: training, calibration schedules, approved settings, work instructions, change control and acceptance rules. Quality control inspects actual batches and files. Both are necessary. A perfect-looking sample cannot compensate for an uncontrolled workflow, and a documented workflow cannot excuse missing or unreadable pages.
质量保证用于设计过程,包括培训、校准计划、获批设置、作业说明、变更控制与验收规则;质量控制用于检查实际批次和文件。两者缺一不可。外观完美的样本不能弥补失控流程,有记录的流程也不能为漏页或不可读页面开脱。
- Completeness: expected containers, items, pages, sides, inserts and attachments reconcile to the manifest.完整性:预期容器、项目、页面、正反面、插页和附件均与清单对账。
- Image integrity: orientation, order, framing, focus, detail, tone, color, scale and absence of introduced artifacts meet the approved specification.图像完整性:方向、顺序、取景、焦点、细节、明暗、色彩、比例及无新增伪影符合获批规格。
- File integrity: extensions match formats, files open and validate, identifiers are unique, metadata relationships resolve and checksums verify.文件完整性:扩展名与格式匹配,文件可打开并通过验证,标识唯一,元数据关系可解析,校验和可核验。
- Usability: target users can find, read, navigate and use the delivered representation without losing material context.可用性:目标用户能够查找、阅读、导航和使用交付结果,同时不丢失重要语境。
Define defect classes and remedies before production. A critical missing page may trigger full batch inspection; a minor cosmetic defect may be accepted. Record the sample method, examined files, findings, rescans, waivers and approver. NARA's Digitization Quality Management Guide announcement links quality assurance, quality control, objective testing and inspection to a managed digitization program.
生产前定义缺陷类别与补救措施。关键漏页可能触发整批全检,轻微外观缺陷则可能被接受。记录抽样方法、被检文件、发现、重扫、豁免与批准人。NARA的数字化质量管理指南公告把质量保证、质量控制、客观测试与检查联系到受管理的数字化项目。
Use InfiniSynapse After Digitized Documents Are Approved数字化文档获批后使用InfiniSynapse
Prepare rights-approved access copies or validated outputs with stable identifiers, versions, permissions, metadata and integrity evidence. InfiniSynapse's public site presents multi-source and multimodal analysis across databases, documents, audio and video. It is relevant when approved digitized documents need analysis with related structured or multimodal sources.
准备权利已获批准的访问副本或验证输出,并确保标识、版本、权限、元数据与完整性证据稳定。InfiniSynapse公开网站展示跨数据库、文档、音频与视频的多源多模态分析能力。当获批数字化文档需要与相关结构化或多模态来源共同分析时,它具有相关性。
Before opening the tool, confirm supported input, authorization, access copy, identifier, version and quality status. Use InfiniSynapse for downstream analysis. Keep inventory, physical handling, capture, OCR, accessibility remediation, metadata cataloging, checksums, preservation and records disposition in their responsible systems.
打开工具前,请确认输入受支持、授权有效,并核实访问副本、标识、版本与质量状态。使用InfiniSynapse进行下游分析;盘点、实物操作、捕获、OCR、可访问性修复、元数据编目、校验和、保存与档案处置仍应保留在各自负责系统中。
Analyze approved digitized documents分析获批数字化文档Review the public InfiniSynapse capability description and verify current source, format, deployment and control support for the intended workload.
使用前请查看InfiniSynapse公开能力说明,并针对预期工作负载验证当前来源、格式、部署与控制支持。
Document Digitization FAQ文档数字化常见问题
What is document digitization?
什么是文档数字化?
Document digitization is the controlled conversion of physical records into digital representations that remain complete, usable and traceable. It includes selection, preparation, image capture, quality control, optional OCR, metadata, packaging, integrity checks, delivery and preservation. Scanning is only the capture step.
文档数字化是把实物记录转换为完整、可用且可追溯的数字再现物的受控过程。它包括选择、准备、图像捕获、质量控制、可选的OCR、元数据、打包、完整性校验、交付与保存。扫描只是其中的捕获步骤。
What is the difference between scanning, OCR and document digitization?
扫描、OCR与文档数字化有什么区别?
Scanning creates a raster image of a physical page. OCR attempts to convert visible characters into machine-readable text. Document digitization is the wider program that governs the originals, capture specifications, image quality, page order, metadata, identifiers, access derivatives, checksums, acceptance and preservation. OCR is optional and must be verified.
扫描创建实物页面的栅格图像;OCR尝试把可见字符转换为机器可读文本;文档数字化则是更广的项目,管理原件、捕获规格、图像质量、页序、元数据、标识符、访问副本、校验和、验收与保存。OCR是可选层,结果必须验证。
How do you digitize paper documents?
如何把纸质文档数字化?
Define the purpose and acceptance criteria; inventory and assess the originals; obtain rights and retention decisions; prepare and order each item; choose capture settings by material and intended use; capture controlled images; inspect completeness and quality; create OCR or access copies if required; attach metadata and checksums; reconcile the batch; then preserve and monitor the approved package.
先定义目的与验收标准,盘点并评估原件,确认权利与保留要求,准备并排序每个项目;再按材料和用途选择捕获设置,受控生成图像,检查完整性与质量,按需创建OCR或访问副本,附加元数据与校验和,对账整批文件,最后保存并持续监测获批交付包。
What resolution should be used for document digitization?
文档数字化应该使用多高的分辨率?
There is no universal resolution for every document. Choose measurable requirements from the smallest significant detail, material type, intended use, capture device, color needs and preservation policy. Test representative samples and validate the actual image performance. A nominal setting by itself does not prove readable, complete or color-accurate capture.
不存在适用于所有文档的通用分辨率。应根据最小重要细节、材料类型、使用目的、捕获设备、色彩需求与保存政策确定可测量要求,并用代表性样本测试实际图像性能。仅有一个标称设置不能证明捕获结果可读、完整或色彩准确。
Which file format is best for digitized documents?
数字化文档最适合使用哪种文件格式?
No single format is best for every role. A preservation master favors fidelity, documented encoding, metadata support and sustainable management; an access derivative favors delivery, search and usability. TIFF, JPEG 2000 and PDF/A may be candidates in particular programs, but the choice must follow current policy, content characteristics, system support, storage and migration plans.
没有一种格式适合所有角色。保存主文件重视保真度、编码文档化、元数据支持与可持续管理;访问副本重视交付、检索与易用性。TIFF、JPEG 2000和PDF/A可在特定项目中成为候选,但选择必须服从当前政策、内容特征、系统支持、存储与迁移计划。
Does OCR make a scanned PDF accessible?
OCR能否让扫描PDF具备可访问性?
OCR can add searchable text, but it does not by itself establish accessibility. Verify recognized text, language, reading order, headings, tables, form controls, alternative text and document structure with appropriate tools and assistive-technology testing. Keep the page image and OCR relationship traceable, and correct material errors.
OCR可以添加可检索文本,但本身不能证明文档可访问。应使用适当工具与辅助技术测试识别文本、语言、阅读顺序、标题、表格、表单控件、替代文本和文档结构。保持页面图像与OCR结果可追溯,并纠正重大错误。
How do you check the quality of digitized documents?
如何检查数字化文档的质量?
Use both process controls and output inspection. Reconcile items and pages, inspect orientation, order, cropping, focus, tone, color, artifacts and readability, validate filenames, formats and metadata, sample OCR against the image, generate and verify checksums, record defects and rescans, and require an acceptance report. Sampling should be risk-based; critical defects may require full inspection.
同时使用过程控制与输出检查:对账项目和页数,检查方向、顺序、裁切、焦点、明暗、色彩、伪影与可读性,验证文件名、格式和元数据,抽检OCR与图像,生成并核验校验和,记录缺陷与重扫,最后形成验收报告。抽样应基于风险,关键缺陷可能要求全检。
Can InfiniSynapse digitize paper documents?
InfiniSynapse能把纸质文档数字化吗?
InfiniSynapse is publicly presented as a multi-source, multimodal analysis tool across databases, documents, audio and video. It can support downstream analysis after digitized documents or approved outputs are ready, but it should not be described as scanner control, a scanning service, OCR, accessibility remediation, a metadata catalog, a preservation repository or records-management software.
InfiniSynapse公开定位为跨数据库、文档、音频和视频的多源多模态分析工具。数字化文档或获批输出准备好后,它可以支持下游分析;但不能把它描述为扫描仪控制、扫描服务、OCR、可访问性修复、元数据目录、保存库或档案管理软件。
Official and First-Party Sources官方与第一方来源
- NARA: Digitization Quality Management GuideNARA:数字化质量管理指南
- NARA: release of the Digitization Quality Management GuideNARA:数字化质量管理指南发布说明
- W3C WAI: OCR on a scanned PDF to provide actual textW3C WAI:对扫描PDF执行OCR以提供实际文本
- Library and Archives Canada: digitization guidelines加拿大图书档案馆:数字化指南
- The National Archives: digitisation guidance英国国家档案馆:数字化指导
- InfiniSynapse: public multi-source and multimodal analysis capabilitiesInfiniSynapse:公开的多源多模态分析能力
These sources establish principles for capture quality, quality management, OCR accessibility, format assessment, authoritative records and digital preservation. Confirm the current version, applicable jurisdiction, record class and institutional policy. This guide's decision framework and hypothetical example are not legal advice, universal technical mandates, customer results or permission to dispose of originals.
这些来源为捕获质量、质量管理、OCR可访问性、格式评估、权威记录与数字保存提供原则。请确认当前版本、适用司法辖区、记录类别与机构政策。本指南的决策框架和假设示例不是法律意见、通用技术强制要求、客户结果或处置原件的许可。
