In a manufacturing setting, vision sensors are used in a variety of control systems, for everything from making sure a robot places a part correctly to inspecting packaged food for quality. A vision sensor uses a camera to capture an image and then a processor to analyze the image, checking a specific aspect of the image for specific traits. It outputs that data to a point in the control system that acts on the data, usually by either allowing the inspected item to proceed in the process or by removing it.

For the examples in this article, a vision sensor checks that fasteners have been inserted correctly into several places on the interior door panel on an automotive assembly line. The sensor verifies that the fasteners are not only present, but fully inserted and straight.

To set up a vision sensor, the operator teaches it what the ideal inspected object looks like. That standard inspected object is called the reference image. In the example of the door panel, the reference image is the component with fasteners inserted at the correct angle and in the correct locations.

To fine-tune the inspection, the operator captures and displays the reference image onscreen and then uses the sensor's software to indicate the following:

• The region of interest -- the area of the image that the sensor should inspect. It can be all or part of the image displayed. In the example, the region of interest is the entire door panel, because the fasteners are set at several points around the component's circumference.


• The features of interest -- the objects to find within the region of interest. In this example, the fasteners are the features of interest.


• The tolerance -- how closely the image being inspected must match the reference image. In the fastener example, the tolerance specifies how precisely the fasters must match the standard location and angle.

The software includes a tool set -- a group of algorithms - that the operator can use to tell the sensor how to locate, identify, count, and measure key features in the image. A sensor can use one or more tools to inspect one or more features.

Vision sensors on the market differ in the tools they include, but the following example gives an idea of the kinds of tools that might be available to the operator, what they do, and the role they play in an inspection.

Depending on how the sensor was configured for the previous inspection, the operator might also modify how the sensor exports data, specifying what data to export, how to delimit it, and whether the sensor connects to a serial port or Ethernet.

After setting up the inspection, the operator can opt to save the inspection settings to use again in the future. After the operator triggers the inspection, the vision sensor extracts the features of interest and analyzes them according to the tolerances. The operator can view the inspection onscreen.