Machine vision is the incorporation of computer vision into assembly and testing automation systems. However, machine vision does differ drastically from computer vision. Computer vision revolves around image processing while machine vision uses digital input and output to manipulate mechanical components. Devices that rely on machine vision are often found at work in product inspection, where they often use digital cameras or other forms of automated vision to perform tasks traditionally performed by human operators. Yet, the way machine vision systems ‘see’ is much different from human vision.
- Digital or analog cameras for acquiring images
- A means of digitizing images, such as a camera interface
- A processor
When these three components are combined into one device, it’s known as a ‘smart camera’. Machine vision systems can consist of a smart camera with the following add-ons:
- Input and output hardware
- Light sources, such as LED illuminators
- Image processing program
- Sensor to detect and trigger image acquisition
- Actuators to sort defective parts
Machine vision systems will work tirelessly performing 100% online inspection, resulting in improved product quality, higher yields, and lower production costs.
To understand how a machine vision system works, it’s helpful to envision it performing one of its typical functions, such as product inspection. First, the sensor detects in a product is present. Once the product is found, the sensor will trigger a camera to capture the image, and a light source to highlight key features. Next, a digitizing device called a frame gabber takes the camera’s image and translates in into digital output, which is then stored in computer memory so it can be manipulated and processed by software.
In order to process an image, computer software must perform several tasks. First, the image is reduced in gradation to a simple black and white format. Next, the image is analyzed by system software to identify defects and proper components based on predetermined criteria. After the image has been analyzed, the product will either pass or fail inspection based on the machine vision system’s findings.