Vision sensor

from Wikipedia, the free encyclopedia

A vision sensor is an image processing system that is optimized for a specific task. The vision sensor takes pictures, evaluates them with image processing algorithms and then triggers a corresponding reaction.

Differentiation from smart cameras

They differ from a smart camera in the following ways:

feature Vision sensor Smart camera
Image processing algorithms Limited, predetermined number Freely programmable
Installation Parameterize or teach-in Program
Frame rate up to 120 frames per second depending on the type, up to 3000 fps possible
Image resolution QVGA - VGA depending on the type, usually more than 1024 × 768 pixels

specialization

Vision sensors are specialized in certain fields of application (industry solution) and are suitable for less complex image processing tasks. They contain interfaces for communication, for example Ethernet, but also industrial-grade 24-volt interfaces for connection to PLC or pneumatic valves .

For reasons of simple operation, the number of hardware and software controls is greatly reduced, so that mostly only minor adjustments to the application are possible.

Components

A vision sensor typically consists of the following components:

  • Integrated lighting mostly light-emitting diodes
  • Integrated lens
  • an imaging sensor, for example a CCD or CMOS sensor and its control
  • a permanent memory for the firmware and the current sensor parameterization
  • an image and data storage mostly as RAM running
  • a real-time image processing processor , implemented as a DSP , CPU , FPGA or as an (embedded) PC
  • Application-specific information processing
  • Interface for connection to an external computer mostly USB , Ethernet - rarely FireWire
  • digital inputs & outputs in 24 V technology for connecting e.g. B. to a PLC

Types

  • One-button sensors with a teach-in button as the only control element (usually for teaching-in a pattern)
  • application-specific sensors with few parameters to be set (e.g. sensors for print image recognition, data matrix code reader), setting using a manual keypad or a remote control unit
  • Universal sensors ( general-purpose sensor ) with user interfaces for extensive parameterization (PC user interface)

The last mentioned universal sensors have the following functions typical for test sequences:

function description application
Gray value comparison checks the percentage brightness Differentiation of light and dark parts
Contrast detection checks the contrast in the picture Attendance check z. B. Lettering on part
Pixel count counts the number of pixels with a certain gray value Differentiation of parts, shape inspection
Contour analysis check the shape Differentiation of parts z. B. Rectangle vs. triangle
Pattern recognition searches for a taught-in pattern in the image Find parts, count them, differentiate between good and bad parts
Distance comparison determines the distance between two gray value edges in pixels Sorting parts of different widths / heights

Web links