OmniDocHow to copy and save changes after converting images to text
After the image to text recognition is completed, how to copy the results, correct typos in the document, and then save it into a file that can be used again. The article gives practices based on common scenarios and lists things to check before and after saving.
Key points of this article
- Don’t copy the recognition results directly. Proofread them paragraph by paragraph against the original image to correct typos, line breaks and punctuation.
- When you need to continue typesetting or saving, export the proofread content to a Word file, which is more stable than just copying the text.
- Clear formatting before pasting into other software to prevent fonts, font sizes and table styles from being carried over.
- Use file names to distinguish versions when saving. The original identification draft and the revised draft are stored separately, making it easier to trace back when errors occur.
- Numbers, names, amounts and dates must be checked item by item with the original image. Machine recognition errors often occur in these places.
Recognizing the text in the picture is only the first step. After completing the recognition, many people found that the paragraphs were broken up, the punctuation was misplaced, and some numbers were misread, so they wanted to copy it directly into the document and change it. As a result, the format was messed up even after changing it, and they had to rearrange it in the end. This article solves the process after identification: how to copy, how to modify, how to save, in order to get a file that can be used directly.
Readers are roughly divided into two categories. One type has only a few sentences on hand and needs to be transferred to chat or email; the other type has full or multi-page materials, such as scanned contracts, handouts or forms, which need to be edited. The two types of people have different approaches, which will be explained separately below. After reading this, you should be able to tell whether you should copy the plain text or export the file, and know where to check before and after saving.
How to copy and save after recognition? One sentence conclusion
After the recognition is completed, proofread it first, and then select the output method according to the purpose: copy a small amount of text directly, export the entire content that needs to be edited to Word, and leave the version with a new file name when saving. The table below can help you choose quickly.
| your situation | Recommended practices | reason |
|---|---|---|
| Just a few sentences to send to chat or email | Copy plain text directly after proofreading | The amount of text is small, just paste it, no need to export the file |
| Full page or multi-page document to continue editing | After recognition, export it to Word and then modify it in Word. | Paragraphs and basic structures are retained more completely, making subsequent editing smoother |
| The content is mainly tables and figures | Check numbers row by row after exporting to Excel | Numerical errors are easier to spot when table columns are aligned |
| The original format needs to be preserved for others to see. | Keep the recognized PDF and attach the proofread text. | The format is consistent with the original, and the transcript is used for reference and citation. |
| Requires multiple people to modify together | Export to Word and open it in a web editor | You can continue to modify it online without sending back attachments. |
Before you start: Confirm these points first
Before copying or exporting, take a minute or two to judge a few things, which can save a lot of rework time.
- Confirm whether the recognition is actually completed. Try to select a piece of text in the PDF or recognition results. If it can be selected and highlighted, it means the text layer already exists; if it cannot be selected, it means it is still a picture and OCR needs to be done first.
- Check the recognition quality against the original image. The clearer and more correct the original image is, the more reliable the recognition result will be. Blurred, tilted, shadowed or shading images will have significantly more recognition errors and a greater amount of changes.
- Think clearly about the end use. Should I send it to others for review, or should I continue to revise it? The purpose determines whether you should export an editable file or just copy the plain text.
- Be aware of content type. The numbers, dates, and names of people in contracts, finances, medical care, etc. must not be wrong and need to be checked item by item; ordinary paragraphs can be read carefully.
- Don’t delete the original picture yet. The original image is the basis for proofreading. It is recommended to keep it together when saving the revised draft, so that you can later trace what a certain word is.
One thing to note here: errors may occur in machine recognition and text modification. Please be sure to check important content with the original image item by item, and do not finalize it after just reading it once.
Steps to save after identifying and modifying in OmniDoc
The following steps take the desktop client as an example. The PDF editor in the client supports OCR on scanned PDFs, can modify paragraph text, and can be exported to Word, Excel or PowerPoint.
- Organize the pictures into a PDF. If the original document contains multiple pictures, first organize them into the same PDF in page order. The page order should be consistent with the original text to avoid reversing the order of paragraphs later.
- Open this PDF in a desktop client. After opening, you can turn through a few pages first to make sure that all pages are there and there are no missing pages or wrong orientations.
- Perform OCR recognition on the scanned page. Perform OCR on the scanned page without text layer. After the recognition is completed, use the "Select Text" method to check again to confirm that the text can be selected.
- Proofread paragraph by paragraph against the original image. Modify paragraph text directly in the PDF editor, focusing on correcting typos, punctuation errors, and numbers. I found that a certain paragraph was split into several lines, and they were merged together in this step.
- Export the Word file after confirming everything is correct. For content that requires more tables, you can export it to Excel and check it separately. After exporting, open it and take a look to see if the paragraphs and titles are basically normal.
- Clear the format before copying to the target location. If you only need text, after copying from Word, select plain text or clear formatting when pasting to avoid fonts, colors and table styles being carried over.
- Save as new file name. Add the date or version number when saving the revised draft, such as "Contract_Proofreading Draft_0301", the original PDF and the file before recognition remain unchanged.
- When collaboration is needed, put it in the web editor and continue to modify it. Open the Word file in UniDoc web editor (app.unidoc.top), you can continue to edit and modify it with others, and then import it back to Word to save it after modification.
Check each item after completion
Once you get the results, don't rush to send them out or file them. Check the following items one by one and deal with the problems found on the right side.
| Check items | How to see | What to do if there is a problem |
|---|---|---|
| Is the text complete? | Compare the original picture and read from the first page to the last page to see if there are any missing paragraphs. | Go back to the PDF to find the missing page, re-execute OCR and then export |
| numbers, amounts, dates | Compare each item with the original picture, especially the decimal point, thousandths and year | Make corrections directly in the revised draft and mark the original drawing for easy review. |
| Paragraph line | Check to see if a sentence is split into multiple lines or paragraphs are merged incorrectly | Manually merge or segment in the editor, then read through it after modification |
| Punctuation and symbols | Pay attention to whether Chinese and English punctuation, quotation marks, brackets and bullets are consistent | Use search and replacement to unify common symbols, and then check a few paragraphs |
| Table structure | Check whether the rows and columns are aligned and whether the merged cells are correct | When the table is complex, export it to Excel for verification, or readjust it in Word |
| save version | Whether the version and date can be seen in the file name, and whether the original is still there | Archive original PDF, revised draft and revised draft separately |
❓ Frequently Asked Questions (FAQ)
1. After the image is converted to text, the format is completely messed up. Can the original layout be retained?
The recognition results can usually only ensure that the text is basically correct, and the original format is difficult to completely restore. If the format is important, it is recommended to keep the recognized PDF as the format version, and save a proofread Word transcript for separate use.
2. The copied text has many extra spaces or garbled characters, what should I do?
First paste into a plain text editor, remove formatting and extra spaces, and then copy to the target location. Garbled characters are mostly caused by the special font of the original image or incorrect recognition language settings. After returning to the original image for confirmation, you can manually correct the corresponding paragraphs.
3. How to save the modified file so that it will not be mixed with the original file?
Save as a new file with "Proofed Draft" and date in the name. Do not overwrite the original PDF and the file before recognition. When modifying multiple times, change the version number each time you save it. If there is a problem, you can go back to the previous version.
4. What aspects should you look for when choosing a free image to text tool?
You can check a few things first: are there any restrictions on the file size and number of pages, whether the exported results are watermarked, whether the files will be uploaded to the server, how to write the privacy terms, and whether the exported format can be edited. These are more likely to determine whether it's a good fit for you than simply looking at "free" material.
5. How much proofreading is required for recognition results to be reliable?
There are no fixed standards. Ordinary paragraphs should be read through, and then numbers and proper nouns should be randomly checked; contracts, financial and medical information should be checked against the original drawings one by one. Errors may occur in machine recognition and text modification. Important content must be reviewed manually.
📝 Summary
The key to converting images to text is not the recognition, but the proofreading, exporting and saving after recognition. First, choose to copy plain text or export to Word according to the purpose. When modifying, check the numbers and paragraphs one by one, and use the version number to distinguish the files when saving, so that the recognition results can be turned into truly usable documents.
