Research And Development Of Fire Detection System Based On Infrared Image

The fire is one of the serious natural disasters.The fire often occurs as a direct result of the annihilation of property and personnel casualties.So, it is emphatic to real-time monitor the fire in the field of fire protection. However,according to the limitations of traditional sense of heat,smoke alarm mode,the methods of image recognition for fire detection is more and more attention,It is an indisputable fact that image is an extension of human vision which can accurately find the fire.Image-based fire detection technology not only can detect fire in the indoor,outdoor and opening up space,but also can achieve early warning and have early protective role by analyzing characteristics of fire images in the early stage,that is according to the early image can trigger a fire alarm when the smoke produced in the early stage of smoke.Fuel combustion release light waves in the frequency range of ultraviolet to infrared,in the scope of visible light,the flame has a unique image of chromatography,texture characteristics,etc.As a result,the image has a clear distinction with the background.In this paper,which is carried out in-depth theoretical study and analysis on fire phenomena.The paper describes the characteristics of fire and flame.It also refers to the basic theory on the infrared radiation and image recognition.The system designs the adaptive threshold algorithm based on Otsu in pretreatment.From the infrared images of fire,analysis of the frame temperature and characterized radiation when the fire broke out in an area by rapid growth,given the rate algorithm,the temperature method,with an area of algorithm and setup the corresponding mathematical models.According to analysis of the acquisition of fire-pixel image for a few seconds continuously,based on the existing pattern recognition clustering method to design a comprehensive algorithm.