Vision System Explained: Industrial Machine Vision for PLCs
An industrial vision system uses cameras, controlled lighting, and image-processing software to inspect, measure, identify, or guide automated equipment. Modern vision systems integrate directly with PLCs over EtherNet/IP, PROFINET, or EtherCAT — no PC required. Used for QA inspection, dimensional gauging, OCR, barcode reading, and robot guidance.
Components of an industrial vision system
- Camera — area scan (still images) or line scan (continuous web). Resolution 0.3 MP to 50+ MP. Mono or colour.
- Lens — chosen for working distance and field of view. Focal length, aperture, mount (C-mount, F-mount).
- Lighting — ring light, dome light, backlight, dark-field, structured light. Often more important than the camera.
- Vision controller — embedded processor running the inspection software. Stand-alone or PC-based.
- Cabling — GigE Vision, USB3 Vision, Camera Link, CoaXPress depending on bandwidth.
- PLC interface — digital I/O for trigger and result, plus EtherNet/IP, PROFINET, Modbus TCP for richer data exchange.
Common applications
- Presence/absence detection — is the cap on the bottle? is the label applied?
- Dimensional gauging — measure hole diameter, part length, tolerance verification.
- OCR / OCV — read date codes, lot numbers, expiration dates.
- Barcode and 2D code reading — DPM (direct part mark), DataMatrix, QR.
- Defect detection — surface scratches, missing components, colour deviations.
- Robot guidance — locate parts on a conveyor, send pick coordinates to robot.
- Assembly verification — confirm components present in correct orientation.
Major vendors
- Cognex In-Sight / DataMan — embedded smart cameras, Cognex VisionPro for PC-based. Industry leader in N. America.
- Keyence CV / IM / IV / XG — strong in Asia, increasingly in N. America. Competitive integrated systems.
- Banner Engineering iVu / VE — simpler point-and-shoot vision for OEM machinery.
- Sick Inspector / Lector — strong in European factory automation.
- Omron FH / FQ-M / V410 — integrates with Sysmac PLCs.
- Basler / Allied Vision / Teledyne — industrial cameras for PC-based vision.
- Beckhoff TwinCAT Vision — vision integrated with PLC runtime in TwinCAT.
Frequently asked questions
What is a vision system in industrial automation?
An industrial vision system uses cameras, controlled lighting, and image-processing software to inspect, measure, identify, or guide automated equipment. Components include camera, lens, lighting, vision controller, and PLC interface. Used for presence detection, dimensional gauging, OCR, barcode reading, defect inspection, and robot guidance.
What is the difference between a smart camera and PC-based vision?
A smart camera (Cognex In-Sight, Keyence CV, Banner iVu) integrates camera, lighting interface, and processor in one device — runs inspection software internally and communicates with PLC via I/O or industrial Ethernet. PC-based vision uses an industrial camera (Basler, Allied Vision) connected to a PC running vision software (Cognex VisionPro, MVTec HALCON). Smart cameras are simpler; PC-based vision handles higher resolution, more complex inspections, and machine learning.
How does a vision system communicate with a PLC?
Three common ways: (1) Digital I/O — trigger from PLC to camera, pass/fail back from camera. Simple but limited to one bit. (2) Industrial Ethernet — EtherNet/IP, PROFINET, Modbus TCP. Camera reports detailed results (positions, measurements, codes) and PLC sends recipe parameters. (3) Vendor-specific buses — Cognex implicit messaging, Sysmac integration. Modern installations use Industrial Ethernet for richer data exchange.
Which vision vendor should I choose?
Match your existing PLC ecosystem. Cognex with Allen-Bradley (native EtherNet/IP integration). Keyence works well with most. Omron FH with Sysmac. Banner iVu for cost-conscious OEM machinery. For PC-based / advanced applications: Cognex VisionPro, MVTec HALCON. Beckhoff TwinCAT Vision integrates directly with PLC runtime for tightly-coupled motion+vision applications.
Why is lighting so important in machine vision?
Lighting determines what the camera can actually see. Wrong lighting hides features (shadows, glare on shiny surfaces) or invents false features (reflections that look like defects). Good lighting design is often more important than camera choice. Common types: ring light (general illumination), dome light (eliminate shadows on shiny parts), backlight (silhouettes for outline detection), dark-field (highlight surface defects), structured light (3D measurement).