A method of operating a computer for smoke and flame detection comprising the steps of: receiving digitised images of the region to be monitored; comparing pixels of one of said images with pixels of another said image according to two predetermined procedures to produce a flame present decision and a smoke present decision; and providing a fire detected signal according to said smoke present and flame present decisions.
In order to allow a smoke detector unit to be utilized in, for example, a domestic kitchen environment, appliances whose operation is commonly associated with the generation of non-hazardous smoke or aerosols are connected to a conventional mains supply socket via a current monitoring unit. The current monitoring units include a radio transceiver, which transmits the operational status of the appliances to a corresponding transceiver of the smoke detector unit 1. If the signals transmitted by the current monitoring units indicate that the appliances and are off, the smoke detector unit generates a warning signal when the smoke density exceeds a lower threshold. If one of the appliances and is detected to be on, the smoke detector unit generates a warning signal only when a second, higher threshold of smoke density is exceeded. Optionally, when the higher smoke density threshold is exceeded; the smoke detector unit transmits the signal to the current monitoring unit to power off the appliances. The invention is also applicable to the detection of other characteristics of ambient fluid, such as temperature or the presence of a particular gas or vapor.
A system and method of tracking moving targets in video images comprises retrieving a reference video frames each comprising arrays of digital pixels; computing a first averaged image for each of the reference video frames, wherein each pixel value of the first averaged image comprises an average pixel intensity value associated with all corresponding pixel values in the reference video frames; computing a second averaged image for a second set of the reference video frames, wherein each pixel value of the second averaged image comprises an average pixel intensity value associated with all corresponding pixel values in the reference video frames; viewing an input video frame comprising arrays of digital pixels; subtracting the input video frame from the first and second averaged images separately to form at two difference images; and combining the difference images to form a change map.
Techniques for determining a characteristic of a flame are provided by the present invention. Provided are methods as well as devices. A flame is monitored across a contiguous spectral range to detect light emitted by the monitored flame. The contiguous spectral range is segmented into a plurality of discrete ranges, and detected light across each of the one or more of the plurality of discrete ranges is respectively processed to determine at least one characteristic of the flame.
An imaging fire detector is used to detect a fire from a recorded image sequence. Either a video camera or an infrared camera is used as an imaging device. Characterizing parameters of a fire are either analyzed individually or in combination. In order to detect changes, an object analysis, a pattern analysis or a spatial-frequency analysis of the recorded image sequences is conducted. Using a threshold value for the duration of the changes, short-term changes are detected and discarded. In a further embodiment, the fire detector may also be used as a burglar alarm. With the aid of communication means, the fire detector transmits a fire alert to a master station.