In typical computer vision systems today, cameras are seen only as simple sensors. The processing is performed after transmitting the complete raw sensor stream via a costly and often distance-limited connection to a centralized processing unit (PC). We think it is more natural to also embed the processing in the camera itself:

What algorithmically belongs to the camera is also physically performed in the camera.

The idea is to compute the information where it becomes available – directly at the sensor – and transmit only results that are on a higher level of abstraction. This represents the emerging trend of self contained and networking capable Smart Cameras.


This approach offers the following benefits:

Low bandwidth requirements

The raw, uncompressed and thus artifacts free video stream is processed directly in the camera. Hence, only the tracking results have to be transmitted which are on a higher level of abstraction.
  • Reduces bandwidth by factor 100+. Hence, Ethernet with virtually unlimited range can be used.
  • Allows the use of standard and inexpensive networks (Ethernet) with virtually unlimited range. Thus, limited cable lengths (i.e., like with CameraLink) are no issue any more.

No additional centralized low level computation necessary

As virtually all of the processing is performed fully distributed within the smart camera network no centralized low level computation of each camera's data is necessary.
  • Server can instead concentrate on higher level (less data intensive) integration algorithms using all smart cameras' outputs as basis.
  • Thus, the servers are no hard bottleneck any more.

Scales ideally with number of cameras

  • The use of intelligent cameras allows that a virtually unlimited number of cameras can be added to the system to keep up with increasing demands in surveillance coverage.
  • Both the network bandwidth and the server resources introduce very soft constraints in terms of maximal number of cameras.
    • SmartSurv allows 100s of 100MBit/s smart cameras on each network port.
    • Using Gigabit Ethernet (GigE), e.g., Matrix Vision mvBlueCOUGARs enables 1000s of cameras per port.

Enables higher resolution and framerate

By using smart cameras, the raw video stream does not need to comply with the camera's output bandwidth any more.
  • Sensors with higher spatial or temporal resolutions can be used, as the raw video stream does not have to be transmitted any more.
    • Matrix Vision offers sensors up to 1600 x 1200.
  • Compression of Intelligent Cameras’ results is of course possible additionally.
    • Compression of raw video within standard IP cameras instead introduces artifacts and requires additional computational resources.

Small, self contained and inexpensive solution

  • The Intelligent Camera approach offers a self-contained vision solution with a small form factor.
  • This increases the reliability and enables the installation at size-limited places.
  • Often more inexpensive than camera & PC solutions, Standard Ethernet instead of CameraLink/FibreLink etc. can be used
  • No High End Server solution is necessary.
  • The low power dissipation of smart cameras enables new mobile applications.

Increased privacy

Due to the fact that the raw video has never to leave each smart camera node, the SmartSurv system can also be applied in privacy sensitive areas. Only the results are then transmitted which are on a more abstract level and not person specific any more.