Image to Excel

A plain OCR tool would read a table as lines of words, one row after another, with nothing to say where one column ends and the next begins. Pasting that into Excel puts the whole row into a single cell. Table recognition adds the step that matters: it works out the structure first, then reads each cell on its own. That is the difference between a sheet you can sort and sum straight away and a block of text you still have to split by hand.

The most useful sources are the ones you cannot copy from: a price list photographed in a shop, a timetable posted as an image, a table in a scanned report, a bank statement saved as a picture, or a screenshot of a dashboard that does not offer an export. Screenshots are the easiest case, since the text is sharp and the lines are perfectly straight. Phone photos work well too when the page is flat, evenly lit and fills most of the frame.

The output format depends on where the data goes next. XLSX keeps headers in bold, merged cells and several tables on separate sheets, and it opens in Excel, Google Sheets, Numbers and LibreOffice. CSV is plain values separated by commas, semicolons or tabs, for importing into accounting software, databases or scripts. HTML table gives markup you can paste into a web page or an email.

Set Table language to the language of the words in the table, such as the column titles and the product or account names. Numbers are read the same way in every language, so a table that is mostly figures comes through well even when the labels are in a script you cannot read yourself. Choosing the right language mainly improves the text cells, which are the ones most likely to need correcting in a photographed table.

When the picture holds running text rather than a grid, such as a letter or a page from a book, Image to Text is the better choice: it keeps paragraphs instead of forcing words into cells. And if the table already exists in a digital file, for example a CSV export, a JSON feed or a PDF made from a spreadsheet, the Data Converter page moves it into Excel without any recognition at all, so every value comes over exactly.

Tables come in two broad styles, and the Table style setting tells the recognizer which one to look for. Most of the time Auto decides correctly. Set it yourself when a result comes back with columns split or merged in the wrong places.

Ruled lines is for tables drawn with visible borders around the cells, as in most spreadsheets printed from Excel, forms and many reports. The lines give an exact map of the grid, so even cells with several lines of text are kept together. Whitespace columns is for tables without borders, where columns are separated only by gaps: bank and card statements, invoices, receipts and many financial reports. Here the recognizer lines up the text and treats steady vertical gaps as column breaks, which is why straight, level photos matter even more for this style.

Two more settings shape the grid. Detect header row treats the first row as column titles, which become bold in Excel and the header line in CSV. Keep merged cells reproduces cells that span several columns or rows; switch it off to repeat the value in every cell instead, which makes the sheet easier to sort and filter.

A few habits make either style read better. Photograph the whole table, including its outer edges and header row, so the recognizer can see where the grid starts and stops. Avoid shooting at an angle, because perspective makes columns lean toward each other and gaps narrow at one side. If a table spans two pages of a document, convert each page and paste the rows together in Excel afterwards, rather than photographing both pages in one shot. When Straighten image is on, which is the default, small tilts are corrected before the grid is found.

The list above is what Store numbers as numbers does, on and off. Leave it on for price lists, statements and anything you want to calculate with. Switch it off for tables full of codes and identifiers, where every character must be kept exactly as printed.

The Decimal separator setting deserves a moment of attention, because countries write numbers differently. In the United States and the United Kingdom, a dot marks decimals and a comma groups thousands. In much of Europe and Latin America it is the other way around. Auto looks at the figures and decides, which works for most tables; set Dot or Comma yourself when a table has few decimals to judge from. For CSV output, a semicolon delimiter and the UTF-8 with BOM encoding help Excel in comma-decimal countries open the file correctly.

Dates need the same care. A date such as 03/04 means the third of April in most of the world and the fourth of March in the United States, and nothing in the picture says which is meant. Check a date with a day above twelve to see how the column was read, and fix the column format in Excel if it went the wrong way. Currency symbols, footnote markers and stray dots next to a figure can also stop a value from being stored as a number; those cells are easy to spot because Excel aligns text to the left and numbers to the right.

Once the sheet is right, the rest of the tools for documents can take it further. Turn the spreadsheet into a PDF to send it to someone who should not edit it, or convert it to CSV, JSON or SQL to load it into another system.