QR Code Recovery Tool

How QR code error correction works

Every QR code contains more data than its payload strictly requires. The extra bytes are Reed-Solomon parity codewords, and they exist for one reason: to let the decoder reconstruct missing or corrupted data without asking the encoder to retransmit. The ISO/IEC 18004 standard defines four error correction levels, each trading capacity for resilience.

The four EC levels

• L (Low): recovers up to 7% of codewords. Best for controlled environments where the code will be printed cleanly and scanned quickly: event tickets, boarding passes, digital screens.
• M (Medium): recovers up to 15% of codewords. A middle setting for product packaging, marketing materials, and general-purpose labels.
• Q (Quartile): recovers up to 25% of codewords. Used when the code is expected to take moderate abuse: warehouse shipping labels, outdoor signage, codes printed on rough surfaces.
• H (High): recovers up to 30% of codewords. Use it when a logo or image is placed over the centre of the code, or when the code will be exposed to harsh physical conditions. The trade-off is a noticeably denser symbol.

Reed-Solomon under the hood

The encoder treats the data payload as a polynomial over GF(256), a finite field with 256 elements that maps neatly onto single bytes. It divides the data polynomial by a generator polynomial and appends the remainder as parity codewords. When the decoder reads the symbol, it evaluates the received polynomial at specific points. If all evaluations come to zero, the data is clean. If not, the non-zero values (called "syndromes") pinpoint both the positions and the magnitudes of the errors. For 2t parity codewords, the decoder can correct up to t corrupted codewords. At level H, a version-5 QR code (37 by 37 modules) carries 46 data codewords and 88 EC codewords in four blocks. Each block has 22 EC codewords and can fix up to 11 corrupted bytes.

Common causes of QR code damage

What damaged the code decides whether recovery is likely and which manual adjustment to try. This tool reads QR codes only. For a scuffed EAN or UPC with digits missing, the barcode repair tool works out the missing digits from the check digit.

Physical wear

Scratches, scuffs, and abrasion from handling, transport, or weather. Warehouse labels that ride a conveyor belt for months lose surface detail gradually. The damage is usually localised, so the error correction can often absorb it. For labels that face this kind of wear, Data Matrix uses the same kind of error correction; you can make one with the Data Matrix generator.

Printing defects

Inkjet bleed, thermal printer fade, toner smudge, or low-DPI output that merges adjacent modules. These defects tend to affect the entire code uniformly, which is harder to recover from than localised damage because every codeword takes a small hit rather than a few codewords taking a large one.

Partial obstruction

A sticker placed over part of the code, a finger in the photo, a fold in the paper, or a receipt that curled and hid one corner. If the obstruction covers less than the code's EC capacity, recovery is straightforward. If it covers a finder pattern, the decoder may fail to locate the code at all, in which case rotation and cropping can help.

Low-resolution capture

Screenshots from video calls, photos taken from across a room, images downscaled by messaging apps. The modules blur together and the decoder cannot tell black from white. The tool's sharpening and contrast enhancement filters target exactly this problem.

Compression artefacts

JPEG compression rounds off the crisp edges that QR codes depend on. Each save-and-share cycle degrades the image further. Screenshots saved as PNG are lossless and decode more reliably; if you have the option, always prefer PNG over JPEG for QR code images.

Tips for better QR code scanning

How you photograph the code often decides whether it scans at all. Try these before you upload a photo here.

Lighting

Even, diffused light produces the best contrast between dark and light modules. Avoid direct flash, which can create glare spots that the decoder reads as white. If you are scanning from a screen, turn the screen brightness up and reduce ambient light to minimise reflections.

Angle and distance

Point the camera straight at the code, as close to perpendicular as you can manage. Steep angles stretch the modules into parallelograms that confuse the decoder's grid alignment. Get close enough that the code fills at least a third of the frame, but leave a margin of quiet zone (the white border) on all sides.

Resolution

Each module in the QR code needs to span at least 3 to 4 pixels in the captured image for reliable decoding. A version-5 code is 37×37 modules, so the code region should be at least 150×150 pixels. Higher is better. If you are cropping a screenshot, do not scale it down below that size.

Focus

Tap to focus on the QR code before capturing. A phone camera may focus on a face or the centre of the frame instead of the code. A crisp image at moderate resolution will decode more reliably than a high-resolution but blurry one.

Frequently asked questions