How can I copy text from a photo and check it for mistakes?
Quick answer
Use a text-recognition feature in a phone, computer, or scanning app to extract the words from the photo, then compare the result with the image before using it. OCR can substitute characters or miss text, especially when an image is blurry, faded, dark, light, ornate, or marked up. Check names and numbers character by character, and review line breaks and punctuation in context. [S1][S2]
The question, the details, the next step.
I have a photo of a printed notice and want to copy its text. What’s the basic process?
Use a text-recognition feature—often called OCR, or optical character recognition—to turn the image of the notice into selectable text. OCR extracts text from an image, but its result is not guaranteed to match the page. [S1] The exact controls depend on the device or app, so look for an option such as “copy text” or “recognize text” in a tool you already use.
After copying, keep the photo visible beside the extracted text. Read both together and correct the copied version before sharing, submitting, or relying on it. Human review matters because recognition can make character substitutions or other errors. [S1]
What should I look at first? I’m most concerned about names and numbers.
Check items where one wrong character can change the meaning:
- Names, initials, and unfamiliar words.
- Phone numbers, dates, reference numbers, prices, and addresses.
- Similar-looking characters, such as 0 and O, 1 and I, or 5 and S.
- Punctuation that changes a value or instruction, such as a decimal point, comma, hyphen, or colon.
Compare each character directly with the image instead of relying on what seems likely from context. For example, if the photo appears to say “Room 10B,” don’t accept “Room IOB” just because the copied text looks plausible. This is an illustrative example, not a result from a particular app. Dark, light, blurry, ornate, or marked-up pages can produce inaccurate recognition, so take extra care with those. [S2]
The words look right, but the copied text has strange line breaks. Should I preserve them?
First decide what the text is for. If you need a faithful transcription, preserve meaningful paragraph breaks, headings, and list structure, and compare them with the image. If you only need the wording for a message or notes, you can join breaks that split a sentence or a word while keeping separate sections distinct.
For example, suppose the photo shows: “Please bring your form and reference number.”
A tidy copy might be “Please bring your form and reference number.” But don’t merge lines automatically: a break may mark a new item, heading, or paragraph. Check the layout in the photo and restore punctuation or spacing only when the image supports it. OCR can produce errors beyond individual letters, so reviewing the structure is part of checking the text. [S1]
Can you show me a complete check using a short example?
Assume, for illustration, that a photo shows this notice: “Collect form A-17 from Desk 3. Return it by 4:30.”
The copied result reads: “Collect form A-I7 from Desk 3. Return it by 4:3O.”
Compare the suspect characters against the photo: the form identifier may be “A-17,” not “A-I7,” and the time may end in a zero, not the letter O. Then check that “Desk 3” and the punctuation agree with the image. If the photo is unclear, don’t silently guess. Mark the uncertain part for review or make a clearer image and check again. This example illustrates a checking method; it is not a claim about a specific device or OCR tool.
A reusable note for uncertain text is: “Copied text: [text]. Please verify [uncertain word, character, or line] against the original photo.”
A practical checklist
Keep the photo visible while checking the extracted text.
Compare names, dates, identifiers, and numbers character by character.
Check punctuation and look-alike characters such as 0/O and 1/I.
Review line breaks, headings, and paragraph boundaries against the image.
Mark unclear text for verification rather than guessing.
A few more useful answers
Does OCR always copy every word correctly?
What if a name or number is too unclear to confirm?
Don’t choose a character based only on what seems plausible. Mark the uncertain part and check the original or ask someone with access to a clearer source to verify it.
Do these steps work the same way in every phone app?
The checking method applies broadly, but the supplied sources do not describe controls for particular phone apps. Consult the instructions for the tool you choose, then review its extracted text against the photo.
Sources
References used for this guide.
- [S1] Creating Digital Documents: OCR and Human Review labs.loc.gov · Checked
- [S2] Library of Congress: OCR Limitations and Thorough Review blogs.loc.gov · Checked