01
Label position drift
Product posture and labeling mechanism changes can create offset, skew, or duplicated labels.
Back to cases
Packaging Case
Beverage Line Labeling Positioning & Label Inspection
A production-line case that connects visual positioning, post-label inspection, rejection linkage, and inspection data retention into an auditable workflow.
Industry
Beverage
High-speed labeling inspection for bottled, canned, and boxed products.
Object
Labels
Presence, offset, skew, wrinkle, damage, duplicate label, and missing label.
Focus
Stability
Line speed, trigger timing, rejection delay, and false reject boundaries.
Pain Points
High-speed labeling lines need stability before anything else
The difficulty comes from speed, material reflection, curved surfaces, space constraints, and rejection timing.
02
Reflective materials
Transparent films, glossy labels, and curved bottles introduce highlights and shadows.
03
Manual sampling gaps
Manual inspection cannot continuously cover every product on a fast line.
04
Rejection timing
Inspection results must stay synchronized with encoder, PLC, and rejection devices.
Vision Solution
Stabilize imaging first, then connect detection and rejection
The project validates optics, trigger timing, algorithms, communication, and rejection logic together on site.
01
Two-station strategy
Pre-label positioning corrects product posture while post-label inspection checks label quality.
02
Lighting and exposure
Lighting, exposure, and camera placement are fixed according to bottle shape and label material.
03
Maintainable rules
Offset, angle, area, edge, and defect thresholds remain configurable for field maintenance.
04
PLC rejection linkage
The system outputs OK/NG and rejection delay while storing images, batch, time, and defect type.
System Setup
A complete station from cameras to data records
Hardware, software, and interfaces are designed together so delivery goes beyond an algorithm demo.
Hardware
01- Industrial camera
- Fixed lens
- Bar / ring light
- Encoder
- Industrial PC
Algorithms
02- ROI localization
- Edge detection
- Template matching
- Offset calculation
- Defect judgment
Line Interfaces
03- PLC communication
- Trigger signal
- Rejection signal
- Alarm output
- Image retention
Inspection Flow
Every product receives a traceable judgment path
The image, decision, rejection action, and data record must remain time-aligned.
01
Trigger image capture
A sensor or encoder triggers the camera when the product reaches the inspection window.
02
Locate label area
The algorithm locates bottle or label references and calculates area, angle, and edge position.
03
Classify defects
Presence, offset, skew, wrinkle, damage, and duplicate label rules output OK or NG.
04
Reject and record
NG results go to the PLC while images, results, batch, and reasons are stored.
Delivery Results
Label quality moves from sampling to product-level inspection
The delivery reduces rework and missed defects while enabling data-based review.
01
Product-level records
Each product keeps inspection result, defect type, and key image evidence.
02
Less manual reinspection
Operators shift from repeated sampling to exception review and equipment maintenance.
03
Verifiable rejection
Test samples and timing records verify that NG products are rejected at the right position.
Reusable Lessons
The reusable value is in optics and acceptance boundaries
Similar projects can reuse sample evaluation, lighting selection, and acceptance sample methods.
01
Classify materials first
Transparent, glossy, matte, and curved labels should be evaluated separately.
02
Build an abnormal sample set
Boundary offset, slight skew, reflection interference, and typical damage should be included.
03
Document parameters
Thresholds, exposure, trigger delay, and rejection delay need clear maintenance notes.
Project Review
Evaluating a labeling inspection or packaging vision project?
Share product photos, label samples, line speed, inspection position, and abnormal samples so we can review feasibility.