How to extract tables from scanned documents

Scans have no text layer. What OCR can recover, and how to give it a fair chance.

Other Doclium tools

All-in-one document tools.
Simple. Fast. Free.

Scan, chat with, and transform your documents. Convert, compress, merge, split and more — free to use.

Drag & drop your PDF here or PDF up to 50MB · or scan with your camera
Try instant demo:

Three workflows a generic PDF converter does not cover.

Popular document tools

Compress, convert, edit and organise files in the formats people use most.

Smart document tools

Extraction, checking and question-answering for when converting a file is not enough.

A scanned document contains no text. It contains a picture of text, which is why you cannot select a word in it and why copying a table out of one gives you nothing. Everything downstream depends on first turning those pixels back into characters. That recovery is imperfect in a specific way worth understanding: OCR does not fail loudly. It does not leave a gap where it was unsure — it produces its best guess, and a wrong guess looks exactly like a right one. An 8 read as a 3 arrives as a perfectly ordinary 3. This is why the useful skill is not finding a better converter. It is knowing which results to check, and giving the scan the best chance before it is read at all.

🔒 Never used for AI training: delete a document and it is gone from the server.
⚡ No watermarks: on any tool, including the free tier.
↩ Undo: step back through your last 10 edits.

How to how to extract tables from scanned documents

  1. 1Scan at a resolution where the smallest digits are clearly legible when you zoom in. Resolution spent on a table is never wasted.
  2. 2Straighten the page before converting. Skew is the single biggest cause of columns collapsing into one.
  3. 3Upload the scan — as PDF or as an image — and let OCR read it, with column structure inferred from the spacing.
  4. 4Check a total. Summing a column and comparing it with the printed total catches most misreads in one step.

Questions people ask

Why do I get one column instead of several?

Skew, almost always. Columns are found from vertical bands of whitespace, and on a tilted scan those bands slant and stop separating the columns. Straighten and try again.

Does a higher-resolution scan always help?

Up to legibility, yes; past it, no. Once the characters are clearly formed, more pixels add file size rather than accuracy. Sharpness matters far more than raw resolution.

Can it read handwriting?

Handwritten figures are the least reliable case by a wide margin, and a handwritten table should be checked in full rather than spot-checked. Printed text is what this is good at.

What about a scanned PDF rather than an image?

Same process. A PDF with no text layer is a picture in a wrapper, so it is read the same way. Running OCR over the PDF first also makes it searchable, which is useful on its own.

Related tools

Explore all tools

NEWCheck a document before you send itFind what is wrong before someone else does.NEWExtract data from a PDFPull totals, dates and references into a table you can use.AIChat with a PDFAsk anything about your document and get answers with verifiable page citations.AISummarize a PDFGet the key points of a long document, each one cited back to its page.AITranslate a PDFRead a document in another language, side by side with the original.AIConvert PDF to MarkdownTurn a PDF into clean Markdown you can paste into docs, wikis or a repo.NEWCompare two documentsSee exactly what changed between two versions.NEWCheck a document before sendingCatch unfilled placeholders, wrong totals, and exposed sensitive data.NEWCheck contract before signingAudit contract clauses, deadlines, signature fields, and hidden liabilities.NEWPrepare documents for applicationVerify document requirements, page limits, and file constraints before submission.NEWCheck PDF for sensitive informationScan documents for exposed credit cards, personal data, and confidential text.NEWCompare contracts for changesHighlight alterations, deleted clauses, and modified wording between two contracts.NEWCheck NDA before signingAudit NDA clauses, term limits, exclusions, and required signature lines.AIAI PDF summarizerGenerate instant executive summaries and bullet points from any PDF document.AIFree ChatGPT for PDF alternativeAsk questions about your documents with exact page citations and zero cloud training.Convert PDF to WordTurn a PDF into a visually exact .docx with one Word page per PDF page.Convert Word to PDFTurn a Word document into a PDF that looks the same everywhere.Convert Excel to PDFTurn a spreadsheet into a PDF anyone can open and print.Convert PDF to ExcelPull tables out of a PDF into a real spreadsheet.Convert bank statement to ExcelExtract transaction tables from bank statements directly into Excel.Extract invoice data from PDFPull totals, line items, dates and references from invoices into a table.Convert image table to ExcelTurn photos and scans of tables into editable Excel spreadsheets with OCR.Convert receipt to CSVExtract totals, tax, and itemized rows from receipts into CSV or Excel.Extract invoice line itemsExtract itemized invoice rows, quantities, and totals into structured tables.Convert PNG to ExcelConvert PNG table screenshots and images directly into Excel (.xlsx).Password protect a PDFAdd a password so the file can't be opened without it.PDF converterConvert to and from PDF — Word, Excel, PowerPoint, images and more.Convert PDF to JPGRender each page as an image you can use anywhere.Convert JPG to PDFTurn photos and images into a single PDF, in the order you choose.Convert HEIC to PDFConvert iPhone HEIC photos into clean, universally shareable PDF pages.Rotate PDF pagesTurn sideways or upside-down pages the right way up.Delete PDF pagesDrop the pages you don't need and keep the rest.Add page numbers to a PDFNumber the pages of any PDF in one step.OCR a scanned PDFRecover the text from a scanned document so it becomes searchable.Redact a PDFRemove sensitive words for good — the text is deleted, not just hidden.Add a watermark to a PDFStamp text across every page — CONFIDENTIAL, DRAFT, or anything you choose.Add a header or footer to a PDFPut a running header or footer on every page, with page numbers and dates filled in.Resize a PDF to A4 or another paper sizeMake every page one consistent paper size, scaled to fit rather than stretched.Highlight text in a PDFMark up every occurrence of a word or phrase in the colour you choose.Sign a PDFDraw or upload your signature and place it exactly where it belongs.Whiteout a PDFCover marks, stamps or stray text with a clean box.Crop a PDFTrim margins and scan edges, on one page or every page.Add an image to a PDFDrop a logo, photo or stamp onto any page.Convert images to PDFCombine photos, screenshots and scans into a single searchable PDF.Convert PowerPoint to PDFTurn a deck into a PDF, one slide per page.Convert HTML to PDFTurn an HTML file into a PDF that prints and shares cleanly.Organize a PDFReorder, rotate and remove pages until the document reads properly.Convert PDF to PowerPointTurn a PDF back into a deck, one page per slide.Repair a damaged PDFRebuild a broken file so it opens again.Fill a PDF formComplete a form and lock the answers before sending.Flatten a PDFFreeze a document so nothing in it can be edited.Remove PDF metadataStrip author names and hidden data that survive redaction.Convert PDF to textGet the words out of a PDF as plain text.Compress PDF to 100KBReduce PDF file size to under 100KB for application and government portals.Compress PDF to 200KBShrink PDF documents to under 200KB while preserving text clarity.Compress PDF to 500KBResize large PDF files to under 500KB for email and web uploads.Remove metadata before sharingStrip author names, software tags, and hidden revision metadata from PDFs.Merge bank statementsCombine multiple monthly bank statements into a single chronological PDF.Redact PDF for court filingPermanently remove sensitive text so it cannot be selected, copied or searched.Flatten PDF form for submissionLock filled form fields into permanent, read-only document pages.Combine tax documents into one PDFMerge tax forms, W-2s, 1099s, and expense receipts into one clean file.Convert scanned handwriting to textExtract readable digital text from scanned handwritten notes and forms.Compress PDF to 50KBReduce PDF size to under 50KB without losing critical document readability.Compress PDF to 300KBTarget exact 300KB file size while keeping text and graphics clear.Compress PDF to 1MBShrink high-resolution PDF documents to under 1MB quickly.Convert CSV to PDFTurn raw comma-separated data into a readable PDF.Convert TXT to PDFTurn a plain text file into a shareable PDF.Convert RTF to PDFConvert rich text to a PDF that looks the same everywhere.Convert ODT to PDFConvert an OpenDocument file to PDF, layout intact.Convert ODS to PDFConvert an OpenDocument spreadsheet to a shareable PDF.Free PDF editor with no watermarkEdit text, annotate, sign, and crop PDFs with clean output and no watermark.Private PDF editorYour documents are never used to train models, and deleting one removes it from the server.Edit text in a PDFFind and replace words throughout a PDF, keeping the original styling.Edit a PDFDouble-click any line to change the text — it keeps the document's own font. Highlight, sign, cover mistakes and crop too.Merge PDF filesCombine several PDFs into a single document, in the order you choose.Split a PDFPull out a page range as its own document. The original stays as it is.Compress a PDFShrink large PDFs by re-encoding images, and see exactly how much you saved.
Free to use No account, no signup, no payment.
Encrypted in transit Your library is private to your device. Privacy policy.
Answers you can check Every AI answer cites the page it came from.
Works everywhere Install it as an app on phone or desktop.
Doclium makes documents simple.

Get in touch

Questions, problems, or a document Doclium handled badly — tell us and a person will read it.

You're Offline
Please check your internet connection and try again.
No document open
Workspace
Open a document to read it here. Ask a question and the cited pages are highlighted.
Ask Doclium
Open a document from Library, then ask it something.
Most used
Document Actions33

Review

Edit

Pages

Convert

Optimize

Protect

Search
Encrypted in transit. Never used to train models.

Doclium

Scan, chat with, and transform your documents.

Popular document tasks

Recent Documents

No document open

AI Actions

Edit & Convert

Organize

Tools

Everything you can do to a document, in one place.

Most used

Doclium Ready

AI

Extract

Convert from PDF

Convert to PDF

Organize

Optimize

Edit

Security

Sign

Profile

Account, storage, and appearance.

?
Not signed in
Free plan
— of —
Help & Support › Privacy policy ›

Upload Document

Drag & drop your file here or Supports: PDF, PNG, JPG, JPEG · Max file size: 50MB

Recent Files

Capture page 1

Drag corners to the page edges
Enhance
Review pages