The idea of machines being able to see and act for us is not a new one. It’s been the stuff of science fiction for decades, and is now very much a reality.
Machine vision came first. This engineering-based system uses existing technologies to mechanically ‘see’ steps along a production line. It helps manufacturers detect flaws in their products before they are packaged, or food distribution companies ensure their foods are correctly labelled, for instance.
Since the development of computer vision, machine vision too is leaping into the future. If we think of machine vision as the body of a system, computer vision is the retina, optic nerve, brain and central nervous system. A machine vision system uses a camera to view an image, computer vision algorithms then process and interpret the image, before instructing other components in the system to act upon that data.
Computer vision can be used alone, without needing to be part of a larger machine system. But a machine vision system doesn’t work without a computer and specific software at its core. This goes way beyond image processing. In computer vision (CV) terms, an image doesn’t even have to be a photo or a video; it could be an ‘image’ from a thermal or infrared sensor, motion detectors or other sources.
Increasingly, computer vision is able to process 3D and moving images as well, including unpredictable observations that earlier iterations of such technology could not handle. Complex operations detect all sorts of features within an image, analyse them and provide rich information about those images.
As computer vision advances, the potential applications for machine vision multiply exponentially. What was once the preserve of heavy industry to determine simple binary actions now appears in the braking systems of autonomous vehicles, compares our faces with our passport photos at airport security gates and helps robots perform surgery.
How machine vision and computer vision work together
Computer vision allows all sorts of computer-controlled machines to work more intelligently and more safely. From large factory and farm equipment, to tiny drones that can recognise a person and follow them automatically, computer vision is helping machines perform better and in more varied ways than ever before.
The merits of machine vision have long been known in heavy industry for inspection purposes. Cameras and computers together can capture and process images far more accurately and quickly than any human. In highly delicate, production line manufacture – such as in creating components for pacemakers – no errors can be made. Human inspectors are simply too risky for such detailed inspections and, when you compare human limitations with the capabilities of a computer eye and brain, it’s easy to see why:
It would take a person ten years to even look at the photos uploaded on Snapchat in the last hour
Many modern manufacturing businesses simply could not remain competitive without computer-driven machine inspections as part of their processes. One of the most common uses is in food production, packaging and distribution.
Machine vision is used every day in reducing waste during the food sorting process, ensuring it is packaged suitably for transport and that all labels are verified. If food is incorrectly labelled, a supermarket will issue an immediate Emergency Product Withdrawal notice (EPW) and issue hefty fines. Too many EPWs can seriously damage the reputation of supplier in an industry that can’t afford to take risks with public health.
With all the information food labels must now contain as a legal requirement, it is simply not possible for a human to inspect the many thousands of labelled items a typical packaging factory produces every day.
What a typical machine vision system looks like
From a structural perspective, below are the standard components of a machine vision system:
- A camera or cameras
- Lighting to ensure the image is clear
- Frame grabber
- A computer and software, for the analysis and processing of images
- Pattern matching and other algorithms may be used, depending on the nature of the images being analysed
- Output components: could be a screen for presenting data plus mechanical components such as a robotic arm, for example
What is the future for vision systems?
There are already so many possibilities for machine vision of the future and those possibilities expand almost daily. As the technology that goes into vision systems advances, the potential for new applications broadens. This is reflected in the growth of the sector. We predict that vision systems will be built increasingly to achieve desired outcomes, rather than existing systems being adapted for new purposes.
New technologies are emerging and being improved all the time. This means machine vision is not only going to become useful for more businesses, it also means the systems that are created will be more flexible and more bespoke for specific needs.
On the computer vision side, deep learning, cloud computing, speedier processors and data integration software are opening up all sorts of opportunities. The factory floor will be able to benefit from machine learning and then share production data with the wider business ERP.
On the machine side, component developments are offering up vastly improved raw materials, such as greater variety of cameras that can be used to create very specific image capture solutions, new lenses, intricate robotics and more.
To learn more about machine vision systems and their enormous business benefits, click below to check out our dedicated page on the topic.