Research Of Burning Material Identification Based On Image Processing

Fire is one of the common disaster in the society,which does great harm to human life and property,and it can cause environmental pollution and nature unbalance.Up to now,the fire automatic alarm system is based on sensors in our country or abroad.But in the circumstance of out-room and large in-room warehouse,the signal of the sensor is so weak that even instruments of high precision can not work due to various of noises.The burning material identification system analyse image signal using computer-seeing and pattern identification,which can identify burning material automatically,and it can report to the person in charge in the moment the fire occurs.In this paper,the attribution and its identification algorithm is deeply studied based on image of infrared radiation.Attribution abstraction is carried out in two ways.one is temperature,which is abstracted in center of gravity.The other is shape,including the extent of roundness and slightness.Second,a method of fire rotation is used to get the slightness of flame.Having processed these attributions,a variety of flames can be identified in the early time,and some interferences can be overcomed.Besieds,the method of pattern identification based on least distance is discussed.In this chapter.Some basic conception is introduced first,then a algorithm based on it is given,finally,some fire samples is tested to prove it.The experiment shows that the system has a good ability of fire identification.In the last,the problem of data trasmition is discussed to strengthen the system.According to the result,buring material identification system based on image processing has a high rate of identification which has a bright future of use.