What to do if the numbers cannot be recognized when converting images to tables: common causes and solutions

Key points of this article

Missing numbers in recognition is usually not caused by the software not recognizing the characters, but by problems with the strokes, lines or resolution of the image.
First check whether each number can be clearly seen in the original picture. Re-taking or straightening the blurred, skewed or reflective photos is more effective than changing tools.
Convert images to PDF and then do OCR. The page size and clarity are more consistent, and the recognition results are easier to check.
After exporting to Excel, you need to check the decimal point, negative sign, comma and date column by column against the original image. These symbols are easily lost or misplaced.

Drag a photo of a table into an online image-to-table conversion website. The table header and text may be recognized, but the numbers are missing one digit, or become spaces, or even recognized as letters. I tried again on several websites, but the problem still exists. Many people encounter this situation when searching for excel tables and unable to recognize numbers, or when converting online images to tables.

This article is written for people who need to convert paper reports, notes or screenshots taken with mobile phones into editable forms. We first explain why numbers are easily missed, then use a table to determine which situation it belongs to, and finally make repairs in the order of easier first and then more difficult. After reading this, you should be able to tell whether the problem lies in the shooting, picture or exporting process, and know which step to change first.

Why are numbers always missing? One sentence conclusion

Numbers are missed mostly because the number itself in the picture is not clear enough and the recognition program does not have enough information to distinguish it. Edit the picture first until each number can be clearly seen, and then do the recognition. The result is usually more stable. The table below provides methods based on your situation.

your situation	Recommended practices	reason
Photos taken with mobile phones are blurry or have shadows	Retake the photo, facing the page, with even lighting	Blurs and shadows will cause digital edges to be lost, making identification impossible.
The table is skewed and the number rows are misplaced.	Rotate and straighten first, then identify	After tilting, the corresponding relationship between rows and columns changes, and numbers are easily linked to adjacent cells.
Screenshot or small font form on the web page	Enlarge and save, or retake the screenshot with a higher resolution	When the strokes are too thin, there are not enough pixels, and numbers such as 6 and 8, 1 and 7 are difficult to distinguish.
Scanned PDF or multiple images	Convert to PDF uniformly and use desktop PDF editor for OCR	The page size and clarity are unified to facilitate centralized inspection of recognition results.
There are many numbers and need to be used accurately	After identification, check column by column against the original image, and manually review important figures.	OCR results may be wrong, and scenarios such as reimbursement and declaration cannot rely solely on recognition results.

First determine what kind of problem it is

Different symptoms correspond to different causes, and the practices are also different. Check the table below to find your situation, and then proceed to the troubleshooting steps in the next section.

symptoms	Common causes	How to confirm
Some numbers are missing or blank	The strokes are broken, the writing is too light, and some parts are covered by reflections	Zoom in on the original image to check if the missing number strokes are continuous and if there are any light spots
Offset the entire row of numbers into adjacent columns	Tables are skewed, or tables have no borders	Use a ruler to compare with the original image and see if the horizontal lines are tilted
Numbers like 6, 8, and 0 are often confused with each other	Low resolution, small font size	Enlarge the original image until every stroke is clearly visible
Decimal points, minus signs, and thousandth commas are lost	These symbols have a small area and disappear first when blurred or compressed	Check cells with decimals and negative numbers separately
The entire page is almost unrecognizable	Entire pages are blurry, underexposed, or just a simple image	Try selecting a passage of text in the PDF; if you can't select it, it means it's an image and needs to be done with OCR
The table is recognized as a segment of text	No borders, unclear layout structure	Check if all the exported content is packed into the same column

Inspect and fix each item one by one

Follow the steps below; the earlier steps solve the problem, and the rest can be skipped.

First, look at the original image. Magnifying it so that you can clearly see every stroke of each number, if even the human eye had to guess it, the recognition program would likely make mistakes too. Retaking blurry photos works better than any setting.
When shooting, face the page directly, avoiding light reflections and hand shadows to ensure even brightness across the page. When shooting with your phone, you can enable grid guide lines to make the horizontal lines of the table parallel to the frame.
Rotate and straighten the skewed image so that the horizontal lines in the table are basically level. The angle only needs to be fine-tuned; do not stretch or deform, otherwise the shape of the numbers will become distorted and may increase misrecognition.
Check if the form has borders. Tables without borders can be filled with vertical and horizontal lines on the image, helping to distinguish columns from rows during recognition. Supplementary lines only need to mark the boundaries of the columns; no changes to the numbers are needed.
Convert all images to PDF. Multiple images can be merged into one file, resulting in more consistent page size and clarity. Then, in the PDF editor of the OmniDoc desktop client (download.omnidoc.top), OCR is performed on the scanned copy to check the recognized text.
After recognition, export to Excel. Before exporting, select a few numbers to see if they can be copied properly; After exporting, compare column by column to the original image, focusing on the decimal point, minus sign, comma, and date. If errors are found, they are corrected directly in Excel and the original image is preserved for review.

How to Prevent It from Happening Next Time

Before shooting, take a test shot to confirm that the small numbers in the form are clearly visible, then take a full page.
When scanning, choose a higher resolution and keep the colors as they are. For text-based tables, grayscale is usually sufficient.
When creating tables, try to keep borders and rarely merge cells; this makes the row and column relationships more stable after recognition.
Uniformly save as PDF, with file names named by date and content for easy access to the originals later.
Don't just look at the recognition results for the sum rows. You can use the sum formula in Excel to calculate it once and then compare it with the sum printed on the original. If the numbers don't match, it means a certain number is incorrect.

❓ Frequently Asked Questions (FAQ)

1. Can table photos taken with a phone be directly used for OCR?

You can try, but if the photo is blurry, skewed, or unevenly lit, missed recognition will increase. It is recommended to first follow the steps above to clear and straighten the photos, then convert them to PDF recognition.

2. If the online image-to-table conversion can't be recognized, is it a problem with the website?

Not necessarily. Common causes include images that are too large, formats not supported, or the image itself is blurry. You can try shrinking the file or converting it to PDF first. When choosing a tool, check its file size limits, whether it supports PDF, and whether it can retain the row-column structure of the table after export.

3. Why are decimal points and minus signs always missing?

These symbols have very small strokes; if the image is slightly blurred or compressed, they disappear first. Reshooting at higher resolution, exporting and then separately checking cells with decimals and negatives usually reveals problems.

4. Can you only recognize part of the table?

Yes, you can. Cropping out the table area before identifying it reduces interference from headers, footers, and descriptions, and makes column relationships clearer.

📝 Summary

Missing numbers is usually due to the image itself being unclear. First, take a clear photo, straighten it, and complete the borders, then convert it uniformly to PDF for OCR. This issue is usually resolved. After exporting to Excel, check decimal points, minus signs, and sums column by column against the original image to ensure data accuracy. If you encounter Excel spreadsheets not recognizing numbers, following this order will help you find the cause faster than repeatedly switching tools.

Open OmniDoc to identify and translate the scan