VCAsys is the culmination of several years of development led by chief executive Geoff Thiel, who established the company in 2007. He previously explored the potential of the technology at hisbusiness PI Vision.
VCAsys software can be embedded into edge devices such as network cameras, uploaded into video management and recording systems, or controlled direct from a PC. VCA will approach all interested video management and network camera manufacturers with a view to integrating with their systems.
VCAsys has been embedded on Texas Instruments (TI) DaVinci DSP semiconductors which are widely used by camera manufacturers to process video images. Leading camera and encoder manufacturer UDP has already deployed VCAPresence in its IPE range of network cameras and NVC range of video encoders.
Two key elements
VCAsys has two key elements: A ‘Tracker’ which spots and follows objects in view, and a ‘Rules Engine’ which allows the user to set rules which then trigger alerts if objects do not conform to those rules.
Two products are included in the VCAsys family from day one: VCAPresence, and VCASurveillance.
VCAPresence offers the facility to detect and track objects and can be used for advanced intruder detection applications in hostile or external environments where early and accurate detection and/or tracking are required.
For tailorable rules setting users will upgrade to VCASurveillance, which is capable of defining objects (such as person, group, animal, vehicle), defining direction and speed of travel, as well as other behavioural indicators like loitering, enter, exit and stopped.
Learns the scene
VCAsys performs ‘learning the scene’ routines to absorb and then adapt to specifics in the scene being analysed. This enables the system to compensate for light and shade in the picture, for example, but also to establish what is moving in a way in which the operator might want to analyse and what is background noise movement (such as trees swaying in the breeze or ripples on water caused by a boat moving through a scene).
Once this six second self-learning routine has been performed, the user can create and shape ‘zones’ and apply rules to these zones within seconds via a drag and drop Graphical User Interface. After this initial learning process there are no further set ups required by the operator for changes in weather or lighting.
A statement said: “Under the bonnet is hundreds of man hours spent developing a unique 3D calibration tool which makes for more accurate analysis of video collected from cameras, particularly those placed high up or directly above a scene.
“This is achieved through adjusting a person icon so that it is the same size as a real person in view. The tool is then able to adjust automatically as the view changes.”
