What is the difference between machine vision and computer vision?

By Donato Montanari, Vice President and General Manager, Machine Vision, Zebra Technologies.

Many supply chain organizations are exploring the potential of more automated and intelligent technologies right now, Zebra Technologies included.

With ever-changing production goals and tighter order fulfilment timelines, manufacturers need easy-to-use solutions that help elevate quality and drive production performance. Warehouse and distribution center operators are looking for ways to streamline the returns process. And retailers are constantly seeking tools that take the burden of increased fulfilment demands off workers, without compromising the quality of the customer experience.

That’s why there’s growing buzz around technologies such as machine vision and computer vision.

But despite their similar-sounding names, these technologies have two very different purposes which, at times, do converge. There are also technologies, such as fixed industrial scanners, that can work in tandem with machine vision and computer vision solutions to bring more breadth, depth and speed to operational visibility and industrial automation. 

It might be easiest to start with the similarities. Machine vision and computer vision are both intelligence-based systems used for image capture, processing, and analysis. In enterprise and industrial environments, their shared value lies in their ability to improve quality control and process control by catching both isolated issues and patterns that a human might miss for whatever reason. For example, both machine vision and computer vision systems are trained to look for discrepancies within some component of an operation. When issues are identified, the systems will notify key stakeholders and then help them decide what steps to take to avoid incurring significant inventory, financial or customer losses.

However, the speed and level at which this intelligence is gathered, distributed and applied is one of the most distinguishing factors between the two types of technologies.

Machine vision is often used on a production line in a manufacturing facility to look for visual inconsistencies in a label, package or even item design that could lead to returns, noncompliance penalties, and other costly consequences. Many use a pass/fail alert structure to help inspectors quickly decide whether items are cleared to continue down the line or should be removed – without having to consult with others first. Machine vision systems also tend to be self-contained, meaning the image capture and analysis occur right there on the line. Data doesn’t have to be sent to a back-office system for processing.

Computer vision, on the other hand, is often used as a back-end processing platform for front-line image capture technologies, such as intelligent automation solutions, bioptic scanners, and even mobile computers. Advanced algorithms are used to help decision makers see what’s happening within their operations and fully understand why it’s happening. Though computer vision still elicits fast decision making and action, there tends to be a little bit longer lead time between the two given the depth and breadth of data being processed through the system. Computer vision is typically a far more comprehensive analysis tool than machine vision, which is much more comparative at a single factor level (i.e., the text on the item’s warning label is supposed to be red but the machine vision camera indicates it is purple.)

In fact, machine vision systems tend to be designed to support very specific industrial automation applications due to their unique line of sight capabilities – and limitations. Computer vision algorithms, on the other hand, can be used more broadly to support qualitative analysis needs. That’s why you’re more likely to see machine vision or some derivative – such as fixed industrial scanning – in manufacturing, warehousing, and distribution environments, and computer vision in retail or healthcare.

Machine vision relies on highly specialised camera technology to reconcile an item or label’s current visual state with what it should look like per the standards guide. However, when someone talks about machine vision, it’s unlikely he or she is referring exclusively to the camera component. ‘Machine vision’ is a cohesive set of technologies and methodologies that are used for the automatic, imaging-based inspection and tracking of work-in-progress items and finished goods from a process control and quality control perspective.

For example, an automotive supplier might use machine vision to improve the speed and accuracy of visual inspections as items move down the line in a discrete manufacturing environment where quality control is critical. The machine vision system – meaning the smart cameras or sensors – positioned either overhead or in line with the conveyor belt can learn to recognise when a part is mislabeled or there is a discrepancy in the design when compared to the blueprints. After capturing a snapshot, an intelligent analysis will occur on the spot to verify the quality of what’s being produced and prepped for shipment. If multiple anomalies are identified by the machine vision system, that could be indicative of a larger process issue that the manufacturer may not have otherwise recognized until after parts were shipped to – and returned by – the customer.

Fixed industrial scanning solutions are utilized more for the track and trace of parts or finished goods as they move down a production, picking, packing, or shipping line, whereas machine vision is used most often for the visual inspection of such items for quality control or process control purposes. That’s because fixed industrial scanners are provisioned to read the barcodes on items moving along conveyor belts or order fulfilment lines in distribution centers and warehouses and provide a status update to logistics managers or possibly even customers. However, fixed industrial scanners – at least the ones Zebra offers – can be used for machine vision with a fairly simple software reconfiguration.

While 88% of manufacturers say they are capturing data about their assets and operations, only 18% believe they are fully equipped to deliver on the Internet of Things (IoT) and connect that data to their business systems to improve operations*. Many warehouse and distribution center operators share the same frustration and challenge. So, Zebra has worked diligently to deliver solutions that help automate the collection, analysis and application of all kinds of data in a simple way.

Zebra’s uniquely designed fixed industrial scanning and machine vision solutions fill a gap in customers’ needs that can’t be addressed as effectively by the other automation solutions. They are the only ones to share the same software platform, which allows a single camera/sensor device to serve a dual purpose, thus reducing an organisation’s investment requirements and streamlining operator training. And they’re both key to improving productivity and efficiency while mitigating quality issues that could challenge an organization’s ability to meet demand.