Self-adaptation of acquisition according to the hour of the day, of the type
of time (sun, not very cloudy, cloudy, stormy), of the position of the building
where is the dangerous area. This self-adaptation results from a measurement
of luminosity which is coupled with the video camera through a processing
software.The software is written in high-level language (C, C++). The user interface (for time of the maintenance actions) was made under the software Visual C++.
The principle of detection is as follows
Detection when a person passes beyond a limit fixed by the software (configurable)
Detection when a person is hidden under materials (brutal absence)
Detection by overpopulation on the zone to be supervised
The images shown above,
are captures of screen, the people (a number and position) do not follow themselves
on these photographs
Temporal filtering of the people detected
with the preceding images
Seek people on the image. This action
has several stages (Adjustment of the color and primary detection of the people,
integration of the image with the preceding, filtering of the false detected
people, stylization of each detected person)
Frame grabbing resulting from a video camera color 