- Home
- Industrial automation
- Machine vision
Machine vision
Machine vision systems and vision inspection.
A camera checks every part, not a sample. We design and integrate a vision system that verifies the completeness, type, position and dimensions of the part, reads the code and stops a faulty part right on the line.

A camera is cheap. Reliable inspection is not.
Machine vision can now reliably replace visual inspection by an operator on the line. Yet the difference between a system that works in a demo and one that sorts parts across three shifts for years is not the camera. It lies in the lighting, the mechanical mounting, the evaluation algorithm and how the inspection is tied into line control. An inspection that keeps rejecting good parts ends up switched off, and one that lets defects through is pointless.
We design vision inspection for quality control from the brief through to commissioning. We test your real parts, including defective ones, choose 2D or 3D imaging and lighting, and tune the evaluation so the system lets no defect through while not rejecting good parts unnecessarily. We connect the result to the PLC, HMI and database. A process engineer can then add a new part type or recipe without a programmer.
Our team has, for example, identified the car body type before painting and compared it with data from the data carrier, error-proofed gearbox assembly at eleven stations of a line, counted pieces in packs using a line laser and a camera, and verified airbag parts against a reference model with archiving to a database. We work with Cognex, Keyence and Omron cameras and the Halcon software library. We can also build quality inspection directly into a robotic cell.
What machine vision checks on the line.
Assembly completeness and correctness
Presence, type and position of components at every assembly step. The system will not let the process continue until the operator has fixed the fault.
OCR and DMC code reading
Reading and verifying DMC codes, barcodes and text on the part or label. The record is matched to the serial number for traceability.
3D inspection and measurement
3D scanners and laser triangulation measure heights, shapes and positions where a flat 2D image is not enough.
Machine vision for robot guidance
The camera determines the position and orientation of the part and the robot picks it without precise fixtures. For randomly placed parts we use 3D vision.
Error-proofing and OK/NOK sorting
The system automatically rejects a faulty part, for example by blowing it off the conveyor, and sends the information to the PLC and to the statistics.
Traceability and archiving
Inspection results, images of faulty parts and parameters are stored in a database. In the event of a customer complaint you can trace the specific part.

What the delivery includes.
We deliver the vision system as a finished function of the line, not as a camera with a licence.
- Feasibility test on your good and defective parts
- Selection of camera, lens, lighting and 2D or 3D imaging
- Mechanical mounting that returns precisely to position after removal
- Programming of the evaluation and recipes for each part type
- Connection to the PLC, HMI, robot and higher-level database
- Operator screen with statistics and the latest NOK parts
- Validation on a production run, commissioning and training
- Documentation, backups and service
Cameras and technologies we work with
How we introduce vision inspection.
Samples and brief
You send us good and defective parts and a description of the defects to be caught. We agree the cycle time and how faulty parts are rejected.
Feasibility test
We verify imaging and lighting on your samples and show you how reliably the inspection works. Only then do we give a fixed quote.
Integration on the line
Installation, connection to the PLC and HMI and recipe set-up. At first we let the inspection only record results and compare them with the operator.
Operation and fine-tuning
After handover we monitor the false reject rate and, on request, add new part types or defect types.
Related services.
3D bin picking
A robot picks randomly placed parts from bins and pallets using 3D vision. Free test visit.
Robotic cells
Turnkey robotic cells: design, safety, grippers, programming and commissioning.
Production data collection
Historisation of data from PLCs and meters, high-availability historian, visualisation, reporting and remote monitoring.
Frequently asked questions.
Did not find your answer? Write to us and we will reply within 24 hours.
How much does vision inspection for quality control cost?
The price depends on the number of inspection points, the camera type, lighting, mechanics and the connection to the line. In our projects, inspection at a single point has typically been in the lower tens of thousands of euros, and a system across several stations of an assembly line in the hundreds of thousands. We give a fixed price after testing on your parts.
What is the difference between 2D and 3D machine vision?
A 2D camera evaluates a flat image, which is enough for checking presence and type, reading codes and most in-plane dimensions. 3D imaging adds height and shape, so we use it for measuring heights and depths and for guiding a robot to randomly placed parts.
How reliable is vision inspection?
It depends on the task, which is why we verify reliability on your parts before we quote. For example, when checking the piece count in packs, the system let no faulty pack through, falsely rejected less than 0.1 % of good ones and ran at 2.5 packs per second.
Can a camera be added to an existing line?
Yes, this is a common case. We add the camera to the existing PLC and HMI, modify their programs and schedule commissioning for shutdowns or breaks between shifts.
Who sets up the inspection for a new part type?
We design the system so that an authorised operator or process engineer can add a new part type or recipe without a programmer. For more complex changes we help via remote access.
Which cameras and software do you use?
Cognex, Keyence and Omron, and the Halcon library for more demanding tasks. We choose the brand by the task, resolution and speed, and by what you already have on site, so your maintenance team does not have to learn yet another system.
Send us a faulty part.
Describe the defect the camera should catch and the line cycle time, and add photos of a good and a bad part. We will get back to you within 24 hours with a proposed solution and a test on samples.
