The so-called smart cameras (or "intelligent cameras"), offer the possibility of being used as a "master" of the complete imaging system.
They are basically artificial vision cameras equipped with an integrated control and image processing system. Therefore, and unlike standard (“dumb”) cameras for machine vision, the use of an external computer or CPU is not necessary. They are also called vision sensors for this reason.
Smart cameras are autonomous vision systems. The user must carry out a small, basic programming in a programming environment that is usually visual and simple. At least the GUI of our smart cameras is very friendly.
The captured image is analyzed in the camera itself based on the parameters of interest defined by the user, according to the specific function or application. This data analysis and processing can then generate actions commanded from the camera to other devices, based on the result of the analysis. For example, expelling the inspected product to one place or another, not taking any action (because everything is going well and correct), extracting the defective part, etc…
For this they have the usual communication interfaces, such as Ethernet connector, I/O ports, or direct communication with PLC.
Logically, the processing speed of smart cameras is (must be) high. A smart camera that was slow would have little chance of surviving.
That said, if the processing speed should be very, very high, then the use of an external computer would be recommended.
Likewise, for very demanding applications, it is usually necessary to develop specific software programs and in this case the use of “dumb” machine vision cameras, without internal processing possibilities.
Many of these vision sensors now include artificial intelligence (AI) and deep learning capabilities, greatly increases their potential, especially in complex and diverse scenarios. The savings in application development time (early implementation, lower engineering cost) and general costs thanks to its high added value mean that the higher price is quickly amortized.
Those that do not implement deep learning or artificial intelligence are now often called “vision sensor” and those that do, “smart camera”.
Typical applications of smart cameras are, among many others:
- as an OCR device (Optical Character Recognition)
- mid-level measurement and metrology
- quality control on finished product
- detection of errors and failures in production
- presence/absence detection
- robot guidance and industrial automation in general
- any artificial vision application with a not very high degree of complication
For code readers only, see the dedicated section.
Families of smart cameras (vision sensors):
|Features||Intro family|| ||- Sorting|
- Target detection
- Code reading
|Max. Resolution||1.6 MP||5 MP||20MP|
|Focal lengths||6-15 mm||6-16 mm||12, 16 mm/x|
|Max. frame rate||60 fps||60 fps||60 fps|
|Existence / Count||Feature matching||Pattern existence / count||y|
|Blob analysis||Spot presence / spot count||y|
|Color||x||Comparison / area||y|
|Deep learning||DL OCR||x||y (GPU)||y (GPU)|
|DL character position||x||x||y (GPU)|
|DL object detection||x||x||y (GPU)|
|DL classification||x||y (GPU)||y (GPU)|
|Defect detection||Anomaly detection||x||x||y|
|Logic tools||Logic||Solution results||y|
|Results judgement||Position fixture||y||y||y|
We are the distributors of HIK Robotics in Portugal and Spain
Other machine (artificial) vision products that might be of your interest:
Area scan cameras Line scan cameras Smart cameras
Zoom / autofocus cameras Polarized cameras 3D cameras
Board cameras Polyspectral cameras SWIR cameras
Line and pattern lasers LED lighting Industrial PCs
Code readers Filters Lenses (fixed f., zoom, telecentric)