Visiopack, the low-cost wireless standalone intelligent video camera

Project partners: Eryma Security System, Martec, Bertin technologies, Varioptic, EADS

The Visiopack project, which was launched by the CEA’s LIST centre as part of PIDEA+ and ended at the end of 2008, was mainly intended to design a video system based on an inexpensive, intelligent, energy-efficient wireless video camera. The video camera’s main advantage lies in its ease of installation, as the systems required to install the unit are fully minimized. Since Visiopack is a standalone unit, it does not require a video network or a mains power supply. As a result, the infrastructure costs slowing the widespread rollout of video surveillance are divided by ten.
The camera is equipped with a solar panel and an infrared sensor. The passive infrared sensor “wakes” the camera when it detects a movement within its field of view. The integrated CMOS video sensor then generates a series of images for analysis by the onboard image recognition software. The software, which can detect whether or not there is a person in the image, generates an alarm if necessary and radios it to a base station.

LIST’s aim was to define the most appropriate computing architecture by ensuring that the average power consumption did not exceed 300 mW. The computational capacities of all onboard processors was studied in detail, and one unit was chosen for its computing capacity and onboard energy management functions (variable frequencies and voltages, and hibernation modes). Lastly, a dedicated image analysis application was developed by LIST’s Onboard Vision Systems team and integrated into the system. Images are analyzed in two phases: in the first phase, processing consists in finding the difference between two consecutive images to find areas of movement. These areas are then analyzed to determine whether the movement is caused by the presence of people or is simply an artefact (caused by the wind, for example).

This outstanding software and hardware system allows LIST to propose the best possible combination of power consumption, computing power and onboard applications.