Research And Design Of Forest Fire Recognition System Based On Image Features

The harm caused by forest fire is devastating and irreversible. Right now, because of forest fire prevention technology in our country is still in the relatively backward stage, the situation is very serious. So there has grown up an urgent need for an effective forest fire monitoring and identification system to solve the problem. The system is based of characteristics of fire smoke and dynamic flame, using the Butter worth high pass filter and fourier transformation function to eliminate the smoke and other impurities generated in the acquisition of forest fire image, so as to improve the accuracy during recognition and extraction of the flame on the dynamic feature. Combined with detailed analysis of the dynamic characteristics of the flame on the extraction of the experimental data, the system can accurately predict whether there is a fire.In this paper, we first describe the development status of forest fire monitoring and recognition technology at home and abroad, and futherly discuss the purpose and significance of this research. Then a detailed description of the components of the forest by WX1700 camera for real-time monitoring and image acquisition and the application of SHT11 as a temperature and humidity sensor chip, and digital output calibration, to ensure that in the process of data acquisition, sensor with high reliability and stability. This method is effective to solve the data acquisition and wireless transmission and data processing problems while utilizing the ARM Cortex-A9 processor to controlle all collection module, and the the Flash memory to store the temperature data, humidity data and image data collected previously through the wireless bridge, which is the foundation for the later software design.We use the method of mathematical morphology to preprocess the image acquisition, removing all kinds of noise and impurity generated in the image acquisition process. In the aspect of smoke processing, we use the method of Butter worth high pass filter for filtering in frequency domain and the Fourier function transformation of eliminating maximizing the image of a smoke, minimizing the impact on the dynamic flame feature recognition and extraction in this paper. In the aspect of flame treatment, we extracte the flame area, flame circularity and flame centroid offset three flame dynamic features to ultimately determine whether there is a fire.The experimental results show that after pretreatment of the elimination of forest fire smoke, then combined with recognition and extraction of dynamic characteristics of the flame, it can effectively improve the flame recognition rate and the accuracy of the identification of fire occurrence. It has a good prospect in forest fire detection technology.