How it works: QR Codes

Friday, May 23, 2025

How it works: QR Codes

QR codes are more than just random-looking dots and squares — they're compact, data-rich barcodes with a sophisticated structure that allows fast and reliable data retrieval, even when partially damaged. Here’s how they actually work:

🔍 What is a QR Code?

A QR (Quick Response) code is a two-dimensional barcode that stores data in a grid of black and white squares. Unlike traditional barcodes (which are 1D and read horizontally), QR codes can store much more information and are readable in 360 degrees.

🧠 Basic Structure of a QR Code

QR codes are not just random patterns — they follow a strict layout. Here's a breakdown of the key components:

Finder Patterns:

The three large squares in the corners (except one) let a scanner locate the code’s position and orientation quickly.

Alignment Pattern:

Smaller square(s) near the bottom right help with distortion correction, especially for larger codes.

Timing Pattern:

Alternating black and white modules that run between the finder patterns help the scanner determine the size and grid alignment.

Version Information:

Identifies the QR code version (size and capacity). There are 40 versions, from small (21×21 modules) to large (177×177).

Format Information:

Stores the error correction level and data mask pattern used.

Data and Error Correction Codewords:

The actual content (text, URL, etc.) is stored here — along with error correction data that allows the QR code to be read even if it's up to 30% damaged.

📡 How Data is Encoded

QR codes store data as binary (0s and 1s), but here's how it's structured:

Text (e.g., a URL) is first converted to binary using ISO 8859-1 or UTF-8 encoding.

Data is split into codewords (8-bit segments).

Reed-Solomon error correction is applied, adding redundancy.

Data and error correction codewords are interleaved and mapped into the QR grid using a zigzag pattern.

🧮 Error Correction

QR codes use Reed-Solomon error correction, which lets them recover lost or obscured data. There are four levels:

L (Low): Recovers 7% of data

M (Medium): 15%

Q (Quartile): 25%

H (High): 30%

This is why QR codes still scan even if they’re scratched, dirty, or partly blocked.

🎭 Data Masking

To avoid patterns that confuse scanners (e.g. large blocks of the same colour), QR codes use masking — applying one of 8 algorithms to ensure the data is evenly distributed. The scanner reads the format information to unmask the data.

🧪 Example: Storing “HELLO”

Convert "HELLO" to binary.

Add metadata (e.g. mode indicator, character count).

Divide into codewords.

Add error correction.

Arrange in the grid with all structure elements.

Mask the data.

Done!

📱 How a Scanner Reads It

Detects finder patterns to orient the image.

Reads format and version info.

Decodes the data and error correction blocks.