Specto

Machine Vision Glossary

Short definitions of terms commonly used in machine vision and camera-based quality control.

We have collected the terms that come up most often in camera-based quality control projects, grouped by topic. Each term is defined briefly; follow the link to the related article or page for details.

Basics

Machine vision
Analysing camera images with software to make decisions about a product. In manufacturing it is used for defect detection, measurement, reading and robot guidance.Learn more
Visual quality control
Quality control based on the appearance of a product. It can be done by hand or automatically with camera systems.Learn more
Camera inspection system
A system of camera, lighting, software and rejector that images the product, evaluates it, removes defective products and records the decision.Learn more
Automated optical inspection (AOI)
Automatic camera inspection of component presence, position and solder joints in electronics assembly.Learn more
OK / NOK
Marking a product as accepted (OK) or rejected (NOK) after inspection. A NOK decision is stored with the reject reason.
False reject
A good product that the system flags as defective and removes. The false reject rate is one of the measures that determines production loss.
Escape
A defective product that the system accepts as good. It is the most critical measure for quality, and acceptance limits are set around it.
Reference sample
Good and defective product samples of known status used to define inspection criteria and validate the system.
Product profile
The set of inspection settings and criteria defined for one product. The matching profile is selected at product changeover.
Region of interest (ROI)
The area of the image where inspection takes place. Examining only the necessary area improves both speed and reliability.
100% inspection
Inspecting every product produced instead of sampling. Camera inspection systems can carry out 100% inspection at line speed.

Camera and lens

Industrial camera
A camera built for production environments that captures images on a trigger signal, with fixed settings and in a repeatable way.Learn more
Area scan camera
A camera that captures the image in one shot as a rectangular frame. Widely used for discrete products.Learn more
Line scan camera
A camera with a single row of pixels that builds the image line by line as the product moves. Used for continuous materials such as fabric, film and sheet.Learn more
Global shutter
A shutter type in which all pixels on the sensor are exposed at the same time. It prevents distortion of moving products.Learn more
Rolling shutter
A shutter type in which sensor rows are exposed one after another. Fast-moving products can appear tilted or distorted.
Resolution
The number of pixels on the sensor. The required resolution is calculated from the field of view and the size of the smallest defect.Learn more
Monochrome camera
A camera that captures greyscale images without colour information. At the same resolution it is sharper and more sensitive to light.
Exposure time
The time during which the sensor collects light. Short exposure prevents motion blur on moving products but needs stronger light.
Motion blur
Blur caused by the product moving during exposure. It is usually kept below one pixel.
Frame rate
The number of images a camera can capture per second. It is chosen for the number of products passing on the line.
Triggering
Capturing the image the moment the product arrives in front of the camera, on a sensor or PLC signal.
Encoder
A sensor that measures the movement of a conveyor or material. With line scan cameras it matches the image to material speed.
GigE Vision
An industrial camera interface standard over Ethernet. It supports long cables and power over the cable with PoE.
USB3 Vision
An industrial camera interface standard over USB 3. It offers a high data rate with short cable lengths.
Focal length
The value that sets a lens’s angle of view. It is chosen from the working distance and the required field of view.
Field of view (FOV)
The size of the area the camera covers in the image.
Working distance
The distance between the front of the lens and the product.
Depth of field
The range of distances over which the image stays sharp. It increases as the aperture is closed, but less light reaches the sensor.
Telecentric lens
A lens that keeps the product’s size in the image constant even when its distance changes. Preferred for precise measurement.Learn more
3D camera
A camera that captures height information of a surface. It works with laser triangulation, structured light or stereo methods.
Laser triangulation
A 3D method that measures height by imaging a laser line projected on the surface from an angle.

Lighting

Bright field
A lighting geometry in which light reflects off the surface into the camera. Flat surfaces look bright; defects that scatter light look dark.Learn more
Dark field
A lighting geometry in which light hits the surface at a very shallow angle. Flat surfaces look dark; scratches and embossing look bright.Learn more
Ring light
A light placed around the camera that illuminates the surface directly.
Dome light
Lighting that spreads light from all directions for a shadow-free, glare-free image. Used on glossy and curved surfaces.Learn more
Coaxial light
Lighting that shines along the camera axis through a half mirror. Used on flat, glossy surfaces.Learn more
Backlight
Lighting from behind the product that creates a sharp silhouette. Used for contours, dimensions and checking inside transparent products.
Strobe lighting
Light that is on only briefly while the image is captured. It freezes moving products and can be driven brighter than continuous light.
Polarising filter
A filter fitted to the light and lens to reduce reflections on glossy surfaces.

Software and AI

Rule-based inspection
Machine vision using fixed thresholds, measurements and rules. Used for clearly defined tasks such as measurement, counting and presence checks.Learn more
Deep learning
An AI method that learns from example images. Used to detect textures, patterns and defects that vary in shape.Learn more
Anomaly detection
Training a model on images of good products only and flagging anything that deviates from them. Suited to rare defects.Learn more
Classification
Assigning an image or defect to one of several predefined classes, for example sorting by defect type.
Segmentation
Assigning every pixel of an image to a class. This is how the location and size of a defect are determined.
Data labelling
Marking the location and type of defects in training images. Model accuracy depends on the accuracy of the labels.
OCR (optical character recognition)
Reading text and numbers in an image. Used to check date codes, lot numbers and serial numbers.Learn more
OCV (optical character verification)
Verifying that the text read matches the expected content and checking print quality.Learn more
DataMatrix and 2D codes
Codes that store data in a two-dimensional square pattern. Used on pharmaceutical packs to carry serial numbers.Learn more
Calibration
Mapping image pixels to real-world dimensions and positions. Needed for measurement and robot guidance.

Quality and production

Rejector
The device that removes defective products from the line. It can work as an air jet, a pusher arm or a diverter gate.
PLC
The programmable controller that runs a production line. A camera system can receive its trigger from the PLC and send the result and reject command back.
Traceability
Being able to trace back from records when, on which line and with what result a product was inspected.
Non-destructive testing
The general name for inspection methods that do not damage the product. Visual testing, X-ray, ultrasonic and penetrant testing are among them.Learn more
4-point system
A fabric grading method in which defects get 1 to 4 points by size and the total is calculated against the inspected area.Learn more
Poka-yoke
An error-proofing method that prevents a defect from occurring or moving to the next step. Camera inspection can act as poka-yoke in assembly.Learn more

Let’s evaluate your project together

Tell us about your product and inspection needs, and we will work out the right system together.