|
Machine vision (MV) is a general term for using software, microprocessor systems and cameras to inspect manufactured products or other inspection criteria in industry. Some computer vision systems may have only a computer or camera but a machine vision system normally integrates into a process where various industrial interface standards can be used.
You can consider Machine vision as a subset of industrial imaging and uses the fields of optics, vision illumination, mechanical , electrical , data communication , software engineering , image processing and machine physics. To the right you can see an image of an Assembly line machine which uses vision to inspect.
If you have an basic inderstanding of these areas, then you have a good starting point in machine vision.
What we have done in this website is try to document the core vision requirements involved when considering any machine system requiring vision.
At the very least you will understand the concepts. Our aim is to keep adding to the site so you can dip in and out as you please, depending on the depth of your knowledge.
The core elements that make up a machine vision system are:
Illumination : Probably the most important aspects of any machine vision project. Lighting is critical. The majority of vision systems are a 2D representation of light reflected from a subject. If the light doesn't reflect properly then the project is off to a bad start.
Optics: Correct field of view and distortion levels are critical. You need to consider the degree of positional accuracy required in a subject that is being inspected repeatedly. Also for measurement applications any distortion introuduced by the camera lens is undesirable.
Cameras: Area Scan, Line Scan, Smart cameras ,firewire , Camera link, etc. The nchoice of camera is dependant primarily on the speed of inspection and the resolution required.
Cabling: How far is your camera form your vision processor?. Digital cameras have restrictions on cable lengths.
Frame Grabbers: Frame Grabber or direct firewire or USB link to your PC. Being designed out of more and more systems but you still need an interface to get your image onto the computer.
Software: Matrox , cognex, your own user defined algorithms. Choose an SDK from a major imaging supplier which will have most of the most common filters and Image analysis routines written for you. But you will still need to integrate them and decide which ones are best for your application
Systems: Combining all of the above into a fully validated system.
The video on the next paragraph, shows a typical application where a machine vision system is used to increase productivity on an assembly line.
|
