"I Can See You", said the Robot

The Difference Between Computer Vision and Machine Vision

I’m far sighted now. Things are blurry when they’re close to my face. I hold small print menus at arm’s length or put my readers on so I can see what salads they have. This is because my hardware has become faulty. However, I’m not end of service as I can get an external hardware upgrade or adjust my form factor with Lasik. For some, like pilots, drivers, and athletes, their vision must be better than perfect. Did you know that machine vision and computer vision allow robots and computers to perform tasks with speed and accuracy that human eyes can’t match?

Let’s differentiate the two. Machine vision is usually a part of a robotic system that zeros in on the most critical thing to make a near perfect decision. Computer vision usually captures a much wider spectrum of vision to extract as much data as possible to identify or correlate what it’s seeing. Combine the two together with a whole bunch of other advanced technologies and you get Spot from Boston Dynamics or?self-driving car systems. In essence these are overlapping technologies in that machine vision is a much more advanced subset of computer vision.

A machine vision system uses a camera to continually view an image or item. Algorithms process the data captured and attempt to determine what is being seen before sending a signal to a software or machine system for action. This is how automated welders make continual perfect welds. Big brother computer vision collects as much data as possible about the entire scene being presented. Machine learning algorithms break the image into components of known things for analysis, validation, data capture, or action. Computer vision can be a standalone technology not connected to a machine such as optical character recognition (OCR) used in intelligent document processing.

So, machines like welders, material handlers, assembly and production lines, and production robots can now see better than a human can. They narrow in on what they’re supposed to see, like a weld line, barcode, or QC spot and do their job expertly based on their vision and programming. Computers can also see at a larger scale and faster than humans can. Images can be broken down and data can be collected, validated, and normalized to help determine software action such as RPA processing an invoice or reading an address on an envelope.

With machine vision and its application in robotics, it’s also important to understand that these advanced technologies prevent human death and injury. Machines that see with machine vision can work in toxic, poorly lit, and dangerous industrial areas. Combining the best of both of these technologies can greatly improve manufacturing capacity throughput while streamlining production operations.

For instance, in a distribution center, a machine vision system can verify that empty available boxes are free of damage and positioned correctly for product placement. It can verify the correct product is put in the correct box and apply a barcode or shipping label. Based on the system software, the machine can then trigger back-end action through machine integration to send certain boxes down specific lines, reroute defective boxes or stop the presses. RPA can also be integrated to perform back-office tasks such as Bill of Material creation, pick and pack analysis, or invoice and PO processing based on machine collected data.

Companies like Comston Robotics, Cognex, KUKA, and Vitronic work to combine robotics, machine vision, and automation to fully streamline manufacturing, accelerate aerospace construction, and innovate food production and distribution. If you’re interested in learning more about the combination of advanced robotics, machine vision, and automation, please reach out to schedule a discovery session.

As my human vision gets worse over time, machine vision and computer vision will continually improve. At least the machines can now help me find my glasses.