SMOKE DETECTION METHOD USING DENSITY DISTRIBUTION CHARACTERISTICS OF SMOKE

摘要

The present invention relates to a smoke detection method using density distribution characteristics of smoke. The smoke detection method using density distribution characteristics of smoke according to the present invention comprises the steps of: a) detecting the movement of an image by receiving the image photographed by a surveillance camera, by a control unit; b) detecting a color of the image received by the control unit; c) determining whether results of the movement and the color of the image are accumulated as many as preset images; d) generating candidate regions of smoke based on accumulated image candidates if the results of the movement and the color of the image are accumulated as many as preset images; e) quantizing the generated candidate regions by using density distribution characteristics of smoke; f) determining smoke detection conditions by analyzing characteristics of smoke and non-smoke in the quantized images; g) detecting a color for a current image frame within the generated candidate regions; h) determining smoke detection conditions by comparing a size between the accumulated candidate smoke regions and the smoke candidate regions after color detection; i) determining whether each of quantization-based and size-based smoke detection conditions is satisfied; and j) finally confirming the detection results as smoke detection by putting the detection results based on two conditions together, if the each of quantization-based and size-based smoke detection conditions is satisfied. Therefore, the smoke detection method using density distribution characteristics of smoke of the present invention enhance accuracy in smoke detection, thereby accurately determining a fire.