Human eyes have been the gold standard in the field of quality control since the concept was first invented.
But relentless pressure to manufacture flawless products has spawned a generation of smart camera technologies with the potential to revolutionise how quality is monitored and documented throughout the plant.
While humans have always played an essential role in visual inspection and scheduled sampling regimes, quality specialists are waking up to the fact that a camera can be programmed to monitor a product or barcode 24 hours a day, and seven days a week.
“When people come to vision for the first time, they know they have a quality problem and they know they could solve it with a camera
Scorpion Vision MD Paul Wilson
The technology has also opened up possibilities in quality assurance that were previously impossible to implement, both in terms of inspection of the food product or packaging, and tracking and traceability, say industry experts.
This new era of quality control has been boosted by the development of off-the-shelf smart cameras, says Mark Williamson, sales and marketing director at Stemmer Imaging.
“With point and click vision systems and improved algorithms, they are now much easier to deploy. Ten years ago it would have been very different.”
While larger companies might have the resources to do their own machine vision integration and development in-house, this new generation of solutions means smaller companies can also implement user-friendly machine vision solutions.
“We sell products to a number of food companies making processing machines for food,” says Williamson.
For example, when you buy a pack of bacon, each one has a very consistent weight and quantity, he says.
“Pigs don’t come as a standard, so a common application is the deployment of a vision system that looks at the front of the bacon being sliced to examine how much fat and lean there is, and then adjust the thickness.”
From there, users can also start to use the solution for grading purposes to get maximum value, he says.
“So while originally you may have a system for quality control, you can end up having it at different stages on the production line. This allows you to identify where the process goes wrong early and prevent wastage,” Williamson says.
It can also help companies in highly regulated sectors such as meat production and pharmaceuticals, to create an audit trail.
“Some of the newer compact vision systems and smart cameras actually have this built in,” says Williamson.
“This means all vision systems are feeding back who logged on, who changed something, so if a product goes out the door that is not compliant, you can see which parameters were used.”
But as promising as the technology is, cameras are not infallible, says Williamson.
“We are still in the situation that a camera is significantly poorer at the odd image that will demonstrate that the batch went through correctly,” he says.
There is a good argument for deploying machine vision technology to help cut costs within the food industry, says Paul Wilson, managing director of Scorpion Vision.
To illustrate, he cites the example of a production line that produces pizzas.
“They will have a metal template designed to measure the pizza to ensure it is the right size. This is done every half hour or so to ensure the products are the right size,” he says.
“If you replace this with machine vision, you are removing the cost of a person doing that task, not very efficiently.”
He adds that this level of automation often brings unexpected benefits too.
“When people come to vision for the first time, they know they have a quality problem and they know they could solve it with a camera. But when they start looking at it more closely, they discover the machine vision system also connects a lot of data,” he says.
Scorpion Vision supplies machine vision software based on a Windows application that is suited to tasks ranging from simple object detection to complex 3D robot guidance.
“We use a 3D camera in oil and gas industries to monitor the position and angle of an umbilical cable being laid underwater,” says Wilson.
“It also goes very well together with organic products, commonly found in the food industry,” he adds.
“You can have 100% quality control and it isn’t subjective, so whether you are measuring a loaf of bread or something more uniform, the same parameter measurements are applied all the time.”
Vision systems can also be used throughout the production process for inspecting the exterior quality of products and packaging, says Dan Rossek, product marketing manager at Omron.
“As a quality control measure, they can be utilised to inspect empty containers, before the actual filling process, or for final exterior quality inspections.”
Because of this flexibility, he says vision systems are generally used for higher value products, for example, verifying serialised 2D codes on labels to fight counterfeits. In a recent Omron project, this feature enabled counterfeit bottles of cheap wine labelled as expensive Italian wine to be successfully detected.
Basler supplies machine vision software for users looking to programme different parameters within cameras such as exposure times or image sequences.
“Process industry customers either want to detect things, sometimes they want to locate things and sometimes they are inspecting and enhancing images,” says René von Fintel, head of product management at Basler.
“It gives them access to everything the camera can do and then this can be integrated into their programme, providing a validation process,” says Fintel.
“The internet of things is about connecting more and more points inside the factory floor through the whole network with automation working together so engineers can monitor the whole process,” he adds.
“With machine vision, the process can be visualised and you get a lot of valuable digital data when cameras are connected to the network.”
Robotics is another field where machine vision is showing great promise, says John Rainer, regional sales manager at Fanuc. The company has been busy integrating 2D and 3D vision sensors called iRVision into its robotic solutions. A robot with vision can perform high-level functions such as picking, inspection, bar code and data matrix recognition, and line tracking, says Rainer. It can also provide an automated solution that is able to mimic the hand-eye coordination of a human and accurately pick random products off a conveyor. “As far as food application goes, we have done a lot of trials and experimentation and in many respects there is no reason our vision can’t be used with food,” he says. It could be applied to the automation of a number of inspection, positioning and orientation tasks, while its code reading capabilities can play a key role in helping food companies to meet product traceability requirements. However, the food industry in the UK is yet to embrace robotics in the way many commentators had predicted, he says.