What is Better for Quality — Visual or Machine Inspection?

Although the human eye is a wonderfully complex instrument, it has its limitations. For instance, when inspecting objects on the production line, human eye cannot compete with machine vision, as the latter is not only faster, but also accurate to a larger extent.

The human eye works in tandem with the brain, allowing us to realize our surroundings. We can recognize things within split seconds, even with their exact shape varying. We can analyze our environment keenly, and although we have a wide field of view under normal conditions, our vision is flexible enough to allow us to focus very sharply on particular areas of interest. As humans learned to survive under different environments and stimuli, the ability of the eye gradually evolved over the millennia.

However, in adapting to our environments and circumstances, our visual capabilities are limited to the natural world. For instance, we have only two eyes, as stereoscopic vision is adequate. We do not need to see moving objects in detail, as the perception of movement is enough. We are sensitive to only a limited portion of the light spectrum, and unable to adjust to glare and reflections, which impede our ability to focus on certain properties of an object for a long time, mainly their size and color. Not only are we quite subjective to perception and storing of images, but our eyes are incapable of making accurate measurements. Therefore, our eyes are not the ideal instruments for verifying product quality.

Automatic inspection and analysis, based on imaging or machine vision, surpasses the performance of the human eye. Machine vision can be more accurate for reliable product inspection, and it is possible to combine it with different technologies to ensure highest quality in production environments.

For instance, in a fast-paced production environment, where long-term reliability is essential, the human eye cannot inspect 20 products moving every second, and where errors detection requires an accuracy of 0.02 square millimeters. Manual inspection with a whole team of people may be attempted, but this would go against the objectivity of inspection. Engineers solve the problem with machine vision. They have six cameras observing the fast moving products. As the cameras use polarized light with strobed exposure and very short shutter speeds, they create extremely sharp images on which the defects stand out perfectly. Computers use special software to search for the defects within 50 milliseconds, and the system can continue to do this for 24 hours a day for every day.

Our eye is capable of learning. We can spot anomalies or defects in products, and recognize it as a defect, although we have never seen the defect before. For instance, we recognize a scratch on a surface as a defect, and know it should not be there. Therefore, we have exceptional interpretative abilities, and for a long time, the human eye was almost unbeatable in this area.

However, machine vision technology is advancing, and in many cases, it is able to rival our interpretative abilities. Although this requires fast computers and self-learning vision algorithms, these are easily available. Therefore, machine vision is catching up fast with the capabilities of the human eye.