Can Laser Code Inspection System

Overview

Production and best-before dates are mandatory markings on food and beverage packaging. Can lines run fast and the codes are engraved on a reflective metal bottom, so manual inspection can neither see every can nor read them reliably:

• Missing or wrong codes are hard to catch: marker dropouts, wrong date settings and broken characters slip through, because manual sampling covers only a tiny share of output.
• Metal can bottoms are hard to read: reflections, concentric ridges and circular text layout make fine laser-engraved characters difficult for ordinary code readers.
• High recall and compliance cost: mislabeled products that leave the plant can lead to batch recalls, regulatory penalties and brand damage.

The system is integrated into the existing conveyor. Cans pass the laser marking, vision inspection and rejection stations in sequence, without changing the line’s takt time. It is running on a beverage production line.

System

Component	Description
Laser marker	Engraves production date, time, batch code and best-before date; no consumables, hard to erase
OCR vision unit	Industrial camera with dedicated lighting images the code area; algorithms locate, read and compare characters
Rejector	Removes unreadable or incorrect cans inline; good cans continue downstream
Control and display terminal	Shows live images and results, records data for traceability and defect statistics

How it works

1. Infeed: cans enter the marking station on the conveyor
2. Laser marking: characters are engraved per the current date and batch rules
3. Image capture: a can in position triggers the camera
4. OCR judgment: the character ring is located, then each character is read and compared with the expected content
5. Reject / pass: failed cans are rejected, good cans continue

Checks

• Presence: detects missing marks, dropouts and missing characters; blank cans are rejected.
• Content: compares against the configured date and batch rules to catch wrong dates and codes.
• Logic: checks that production date and best-before date match, avoiding parameter errors.
• Quality: detects blurred, broken or offset characters.

Defects detected on site

The defect images in the gallery were all captured on the production line and judged NG by the system, then removed by the rejector.

Defect	Description
Upside-down	The can enters inverted, so the camera sees the lid with the tab. The system recognizes orientation and rejects it
Missing characters	Only part of the code is printed. The character count is compared with the set value, so even one missing character is caught
Double printing	The can is marked twice; overlapping characters are detected as extras
No printing	No characters on the bottom, usually caused by a marker that did not fire; character count is 0
Fallen can	The can has fallen over and no normal bottom image is captured; it is rejected

Software

The software shows the current image, character count and result (OK / NG with defect name) in real time. NG images are saved by time and can be reviewed under defect browsing. Good and NG counts are summarized by shift and batch, and parameters such as expected character count can be changed directly on screen.

The same vision architecture can be adapted with different cameras, lighting and rules for bottled drinks, canned food, personal care and pharmaceutical packaging, and carton codes.