The Software Design For Forest Fire Monitoring Based On Embedded Linux

A single image characteristic is analyzed based on traditional forest fire detection algorithm which is designed only for smoke or flame fire.The final test results of accuracy is always affected by the light mutation and the rustling leaves.So the forest fire detection algorithm based on video image can be improved on many affects.In fact,in the early stage of the forest fire, more smoke is produced which can be generated as a forest fire important signal.So a algorithm that combination of smoke and flame of the forest fire designed is of great significance.With studying the traditional forest fire detection algorithms based on the image characteristics and the flame features of unrules throbbing,diffusivity,contour,edge smoke image information and flap, the forest fire detection software based on embedded linux is designed and implemented.The fire smoke is first detected.The system make a simple judgment according to the difference between the foreground image which is extracted using background based on the dynamic threshold and the reference image,if the difference value is greater than the threshold setted,then the image which is cached is analyzed using wavelet transform,if the result of the wavelet transform are within the values setted,then make a further judgment according to the diffusivity of image.The forest fire flame is second detecte.First of all,the traditional gaussian mixture background difference method is improved to be adapted to the influence of illumination mutaion and trees swaying in the forest which can detect the moving targets effectivly. To improve the computing spped of algorithm,some of the initialization parameters are optimized.The motion area which is extracted is analyzed according to the color information of fire flame and the characteristics of random changes in the agreed upon period of time.Dynamic detection algorithm based on gaussion mixture model and static detection algorithm based on color and contour information is combined to improved the accuracy of the fire flame judgment.Finally, the fire smoke algorithm and flame composite detection algorithm is transplanted to the TMS320DM8148 and the experiment results show that the algorithm proposed in this artical has a advantages of high accuracy and reliability.