Blurry QR Code Reader: Decode a Damaged or Unclear Code

How to read a blurry QR code

1. Give it the image: drop the photo or screenshot on the box, tap to browse, or press Ctrl+V to paste from your clipboard. Nothing is uploaded, it is all read in your browser.
2. Watch the attempts: the tool works through nine processed versions of your image in order, cheapest first, and stops the moment one of them reads.
3. See what worked: the successful attempt is named, and if it was not the original you get to see the processed version that did the job.
4. Read the result: the decoded content appears in full, with a copy button and, for links, a button to open it.
5. If nothing reads: you get a specific diagnosis of what is wrong with that particular image, measured from it rather than guessed.

Why a QR code blurry enough to fail still might be readable

Whether you would describe it as a QR code blurred by a shaky hand, a blurry QR code from a screenshot, or simply a damaged QR code that no scanner will touch, the question is the same: is the information still in the file, or has it gone? The answer decides everything, and it is worth understanding before you waste time on it.

A QR code stores information in the size and position of its black and white squares. A decoder finds the three large squares in the corners, works out the grid from them, then reads every cell. That whole process depends on neighbouring squares staying distinguishable from each other.

Blur destroys exactly that. Once two adjacent squares have smeared into one grey patch, the file no longer contains the information about which was which, and nothing can recover it. This is worth being straightforward about: in testing, a code hit with a moderate 3 pixel blur could not be read by any of nine processing attempts, and no tool anywhere will do better.

What often can be saved is a code that is intact but hard to see. Those are different problems with different fixes:

• Too dark, too pale or washed out. The squares are all still there, just compressed into a narrow band of greys. Stretching the contrast pulls them apart again.
• A shadow across part of it. One global brightness cutoff cannot suit both the lit and shaded halves. A local threshold judges each area against its own surroundings instead.
• Inverted colours. Light squares on a dark background, common in dark mode screenshots and in styled codes. Some readers only look for the normal arrangement and see nothing at all, though the decoder here checks both.
• Too small, or mangled by compression. Enlarging adds no detail, but it can give the decoder enough pixels per square to lock on, and a clean black and white conversion strips away compression noise.

What about a blurred barcode?

A blurry barcode is a harder problem than a blurry QR code, and the reason is worth knowing. The ordinary striped barcode on a product, a Code 128 or an EAN-13, carries no error correction at all. It has a single check digit, which is just enough to notice that something is wrong, and nothing whatsoever to repair it with. A QR code by contrast can lose up to about 30 percent of itself and still rebuild the missing parts.

So a blurred barcode that will not scan is usually finished, where an equally blurred QR code often is not. If yours is a striped product barcode rather than a square QR code, our barcode scanner is the right tool, and it will also tell you whether the number it reads passes its check digit, which catches a misread that looks plausible.

What the nine attempts actually do

They run in order of cost, so an easy image is decoded almost instantly and only a difficult one works through the whole sequence.

1. Original. Many codes that look bad to a human are perfectly readable to a decoder.
2. Contrast stretched. The darkest and lightest parts are pushed out to true black and white.
3. Black and white by Otsu. Picks the single cutoff that best separates the two groups of brightness in the image, rather than assuming the middle.
4. Local threshold. Each pixel is judged against the average of its own neighbourhood, which is what handles a shadow falling across the code.
5. Local threshold, wider window. The same idea over a larger area, for gentler lighting changes.
6. Sharpened. Edges between squares are pushed back apart before the cutoff is applied.
7. Inverted. The whole image is flipped, for light-on-dark codes.
8. Enlarged twice. More pixels per square, which sometimes is all the decoder needed.
9. Enlarged three times. A last attempt for very small crops.

The tool tells you which one worked, which is genuinely useful information. Measured on ten deliberately damaged codes, an unprocessed decode failed on four of them and the sequence recovered three: two needed nothing more than contrast stretching, one needed a clean black and white conversion, and one was destroyed past recovery. If stretching the contrast is what rescued your code, the original is simply too flat, and rephotographing it in better light will fix it properly.

How much damage a QR code is built to survive

More than most people expect. Every QR code carries Reed-Solomon error correction at one of four levels, recovering roughly 7, 15, 25 or 30 percent of the symbol depending on the level the creator chose. That is why a code with a logo printed over the middle still scans, and why a torn corner often does not matter.

It also sets the limit. If a code will not read at all, the damage has usually gone well past what its correction level can absorb. Our guide to QR code error correction levels covers how to pick a level that survives real printing.

Is this a blurry QR code scanner?

Not in the live-camera sense, and deliberately so. If your phone camera could read the code you would not be here, so pointing another camera at it changes nothing. This works from a still image instead, which is what makes the extra processing possible. On a phone you can still take the photo from this page: the shutter button above opens your camera, and the picture goes straight into the same nine attempts.

If it still will not read

Stop processing the photograph and go back to the source. The QR code was only ever a wrapper around something someone could give you directly, so ask the sender for the file, the link or the details themselves.

If it is on something printed, take the photo again rather than working on the one you have. Fill the frame with the code, hold the camera square to it rather than at an angle, use even light, and turn the flash off, since flash usually blows out the middle of a glossy printed code. One good photograph beats any amount of processing on a bad one.

If your code is fine and the problem is that you have no camera, use our QR code reader for images and screenshots instead, which is the simpler tool for a clean code. For a damaged product barcode rather than a QR code, try the barcode scanner, which also checks whether the number it reads is valid. And if your codes keep coming out unreadable when printed, why QR codes come out blurry covers the causes at the printing end.

Testing a design before it becomes permanent? Print it, photograph it badly, and read it here. If a degraded photo still decodes, the design has margin. Planning a QR code tattoo covers the size it needs before you get that far.