An Introduction to Machine Vision
What is Machine Vision Inspection?
Machine Vision Inspection is a technology based on image processing that is used to automate inspection processes in production lines in various manufacturing industries. Basically, it allows a computer to inspect and evaluate product quality using industrial cameras, lights, and computer vision software.
The technology is used to perform high-speed, reliable quality control to ensure product quality and improve overall efficiency as it releases manual labour from repetitive and challenging inspection tasks and/or expands the existing quality control.
Additionally, by digitising the inspection process and generating documenting data, machine vision inspection can be used to provide production data and statistics, which can facilitate production optimisation.
Due to its wide application and efficiency, machine vision inspection is widely and increasingly used in several manufacturing industries such as automotive, electronics, food and beverage, pharmaceuticals, packaging, labelling, and printing.
Machine vision-based inspection is carried out by a machine vision system.
What does a Machine Vision System consist of?
A Machine Vision Inspection System consists of hardware components, the system’s body, and computer vision software, which can be considered as the system’s brain.
The Hardware Components
A Machine Vision system’s exact hardware setup depends on its specific inspection tasks and hence configuration, but basic components of a Machine Vision System count:
- Sensors that identify the presence of incoming objects
- Cameras, the eyes of the system, used for acquiring images of the
objects that are to be inspected.
- Lights and often a housing to ensure optimal light for the image acquisition.
- A PLC for controlling the system logic i.e., inputs, outputs and other actions from hardware components.
- An industrial computer running the computer vision software that analyses the acquired images and provides a HMI.
- An operator panel that visualises data and enables operators to control the system.
The Computer Vision Software
The brain of the Machine Vision System, the computer vision software, should be perceived as a library containing several software packages that each provides a specific inspection type, i.e., enables the system to inspect and evaluate specific elements of an object.
It processes and evaluates the acquired image of the inspected object to answer the important question: Does the inspected object have any defects compared to the desired quality level set by the producer?
How does Machine Vision Inspection work?
When a machine vision system inspects an object, the following steps take place:
- Image Acquisition
First, when the incoming object approaches the system, it is detected by a sensor. Light sources and cameras act on the sensor’s input by lighting up the area and capturing an image of the object at exactly the right time.
Nothing happens without a successful image acquisition. Using the right optics and lighting technique is key to ensure high image quality, which is fundamental for carrying out an accurate, reliable analysis and hence quality control.
- Image Processing and Analysis
The acquired image is pre-processed by applying algorithms and a number of specific filters – to detect the product in the image and highlight the key features of interest. Based, on the filtered image, algorithms are then used to generate a data object – data containing the characteristics of inspected object’s highlighted features.
The design of an algorithm impacts the efficiency of the image processing and analysis. Powerful algorithms facilitate accurate, high-speed vision inspection, and hence maximises product quality and throughput.
- Result evaluation
The system’s software compares the extracted data object, i.e. the characteristics of the object’s highlighted key features with a set of predefined tolerances, which defines a producer’s target quality level.
- Decision Making
Finally, based on the comparison between the processing results and the set tolerances, decisions are made by the system. The inspected object will get a classification e.g. passed or failed. Some machine vision systems will be able to reject the product or initiate warnings, alarms, etc.
- Process Control
Some machine vision systems additionally utilise the gathered inspection data to spot trends in the production, and give recommendations to the machine operators for production optimisation. This can be done via data platforms that provide production data overviews highlighting crucial production information.
An overview of production data can help manufacturers to understand why faulty products and errors occur and support them in identifying root causes. Also the adjustment of process parameters can be monitored and evaluated to optimise the overall production, and the performance of different shifts, production lines and production facilities can be compared to align and optimise production procedures.
Machine Vision Inspection in the Food Industry
Machine Vision inspection is an attractive technology for food producers to meet the high demands and cope with the challenges they face.
Not only are customers and retailers increasingly demanding high product quality and quality documentation – food producers simultaneously need to optimise to remain profitable, focus on sustainability and deal with labour shortages.
What can Machine Vision Systems check?
The application of machine vision inspection is vast, which means that any food producer in any sector can utilise the power of the technology to handle specific issues related to product quality.
Machine vision systems can be implemented into different production lines and configured to match specific application and inspection requirements.
Food producers can utilise machine vision inspection to perform:
- Inspection of food and packaging appearance
- Presence/absence check
- Orientation check
- Verification of amount and shape
- Colour inspection
- Foreign object detection
- Seal inspection and validation
- Label check and verification
- Reading and check of printed date markings
- Reading and check of barcodes and QR codes.
Machine Vision Systems are The Eyes of The Robots
As automation is key to increase efficiency and deal with labour shortages, producers are focusing on automating older existing production lines. Robots are replacing people in processing and packaging lines to increase throughput. And new production facilities are built fully automated today.
Hence, dark factories are the future, meaning that factories in theory would require no lights, as all processes are fully automated by machines.
It also means that new quality control solutions are required as producers do not have the same number of people on the production floor to monitor product quality, and increased production speeds due to an increasing automation level leave workers with inspection tasks that challenge the human eye.
This makes machine vision inspection a technology of the future. With its ability to inspect all produced products and detect the tiniest deviations and errors consistently and reliably over time at high speeds, enabling producers to meet the high demands of consumers and retailers while increasing efficiency.
More Machine Vision Insights
How to utilise the Machine Vision Technology
Even though machine vision holds great potential for boosting quality and production performance, it is a technology that is difficult to master and utilise consistently for efficient quality control. But if utilised properly, machine vision can boost a producer’s quality and production performance.
Read about how to achieve efficient quality control by using machine vision here
Discover all the benefits of Machine Vision Inspection
If utilised properly, food producers can implement machine vision systems in both food processing and packaging lines to improve product quality and documentation while reducing costs and facilitating optimisation of production processes.
Discover how Vision inspection in many ways can improve food producers’ business performance here.