#machinevisionintegrator

What is Machine Vision (MV) ?

Machine vision (MV) is the technology and methods used to provide imaging-based automatic inspection and analysis for such applications as automatic inspection, process control, and robot guidance, usually in industry. Machine vision is a term encompassing a large number of technologies, software and hardware products, integrated systems, actions, methods, and expertise.

Example 1

Piston Ring Counting – This machine is used to count piston rings and can count different models of rings ranging from a minimum thickness of 0.25mm.

(Picture credits – Qualitas Technologies)

Problems that are most likely to occur – To pack the stack of rings, the counting of the piston rings has to be done. And it is a tedious and time-consuming process. Also, the accuracy for lesser thickness can go down due to human errors.

Example 2

Gear teeth counting machine – This machine is used to count the number of teeth available on the machine gears and classify the gears based on the number

(Picture credits – Qualitas Technologies)

Problems that are most likely to occur – Counting the teeth of gears is highly essential because of it’s vital role in generating the required torque, but the diameter of the gears and patterns of the teeth varied over a wide range based on shape, teeth height, thickness etc and counting it is a challenging task.

Why Machine Vision?

While human inspectors working on assembly lines visually inspect parts to judge the quality of workmanship, machine vision systems use cameras and image processing software to perform similar inspections.  Machine Vision inspection plays an important role in achieving 100% quality control in manufacturing, reducing costs and ensuring a high level of customer satisfaction. Machine vision system inspection consists of narrowly defined tasks such as counting objects on a conveyor, reading serial numbers, and searching for surface defects. Manufacturers often prefer machine vision systems for visual inspections that require high speed, high magnification, around-the-clock operation, and/or repeatability of measurements.

Few other advantages of using Machine vision –

Accuracy – Today’s machine vision systems have a high degree of accuracy that can be achieved. With advances in learning as well as artificial intelligence you could actually build machines that can surpass human accuracy.

Reliability – This is another major advantage of Machine vision. Humans aren’t really designed for repetitive tasks. We are creative in nature. If you put a factory worker in assembly line and ask him to do the same thing over and over again for like 12 hours, he cannot be relied upon for giving accurate results. This won’t happen with Machine vision.

Inspection of the “invisible” – The human sight is limited to what’s in the visible spectrum. And that’s typically 400 to 700 nanometers. But with advanced multi spectral, hyper spectral imaging systems you could actually go beyond these ranges, see things which are not visible with the naked eye. Common applications of multi spectral imaging could be in food processing, health care, and pharmaceutical or even the military.                                                                                                                                                                                                                                          

Can it really replace humans?

Machine vision systems have made huge leaps in innovation in the past decade or two alone.  They’re used in everything from traffic and security cameras to food inspection and medical imaging – even the checkout counter at the grocery store uses a vision system!

When we look at each sub-component (ex: camera and Software), there’s no doubt that machines outperform humans.

Cameras

There are much faster cameras, they can reliably and with much higher precision capture images just not comparable to the human eye. HS and MS cameras can image scenes which are outside the visible spectral range.

Difference between human eye and camera

ANGLE OF VIEW

With cameras, this is determined by the focal length of the lens (along with the sensor size of the camera). For example, a telephoto lens has a longer focal length than a standard portrait lens, and thus encompasses a narrower angle of view.                                                                                                        Unfortunately our eyes aren’t as straightforward. Although the human eye has a focal length of approximately 22 mm, this is misleading because

(i) the back of our eyes are curved,

(ii) the periphery of our visual field contains progressively less detail than the center, and

(iii) the scene we perceive is the combined result of both eyes.

RESOLUTION & DETAIL  

Most current digital cameras have 5-20 megapixels, which is often cited as falling far short of our own visual system. This is based on the fact that at 20/20 vision, the human eye is able to resolve the equivalent of a 52 megapixel camera (assuming a 60° angle of view).

However, such calculations are misleading. Only our central vision is 20/20, so we never actually resolve that much detail in a single glance. Away from the center, our visual ability decreases dramatically, such that by just 20° off-center our eyes resolve only one-tenth as much detail. At the periphery, we only detect large-scale contrast and minimal color.

Software

This is highly consistent for repetitive tasks and don’t fall prey to fatigue or boredom issues, etc. They are also consistent in decision making.For example, give 1000 images to a human at different days or times , the results will vary due to various factors and there is no consistency here. But the software will always give consistent results.Deep Learning is gaining much popularity due to its supremacy in terms of accuracy when trained with huge amounts of data.                                              Practically, Deep Learning is a subset of Machine Learning that achieves great power and flexibility by learning to represent the world as nested hierarchy of concepts, with each concept defined in relation to simpler concepts, and more abstract representations computed in terms of less abstract ones.For example,Language recognitionDeep learning machines are beginning to differentiate dialects of a language. A machine decides that someone is speaking English and then engages an AI that is learning to tell the differences between dialects. Once the dialect is determined, another AI will step in that specializes in that particular dialect. All of this happens without involvement from a human.Image caption generationAnother impressive capability of deep learning is to identify an image and create a coherent caption with proper sentence structure for that image just like a human would write.

However, When It Comes To The System As A Whole, The Human Capability Is Still Largely Superior.

Multi-tasking – Humans can work on multiple responsibilities unlike machine vision where in the time required to teach system on each and everything is considerably high.

Decision making – Humans have the ability to make decisions from their past experience. But, even the most advanced robots can hardly compete with a 6 years old kid.

Augment Not Replace!

AI over the next few years only automates tasks, within broader processes, that are currently handled exclusively by humans. Organizations will divide many of their critical processes into a series of smaller tasks and see where they can benefit the most from automation and which tasks need to remain with humans. The goal here won’t be to displace people but to use AI to augment existing processes.

Machine Vision Is Reactive In Nature. It Only Tells You When Something Is Wrong Or Has A Defect.

For example, Finding the defects on the surface of gun parts. As this is a special case of analyzing the surface defects due to the visibility of defects only under UV light, the image acquisition was done using a color camera and UV light in the factory condition. The defects were clearly visible and trained accordingly.

(Picture credits – Qualitas Technologies)

Machine Vision can be used to segregate sure defects and unsure defects. Only unsure defects can be re-verified by humans.

One such example is, usage of Machine vision in defect detection.

Machine vision is used to detect surface defects on the UBS line (Under-body sealant) which is hard to inspect continuously by a human. Hence, AI based Machine vision is used here to do the task effectively and when a defect is identified, human inter-vision is needed re-verify the detected defect and fix it. This way humans and Machine vision technology join hands which results in augmentation.

Manual quality control to sample the output of machine vision systems identify gaps and errors.

An ideal example for this would be,

Online reading of QR code and characters on Blisters which was soporific in nature and most importantly less accurate.

(Picture credits – Qualitas Technologies)

Conclusion