Research On Key Techniques For Infrared Imaging Forest Fire Prevention

Forest fire prevention is one of the important measures to protect forest resources,improve forest coverage and improve the quality of human living environment.With the development of computer technology,monitoring of forest fires by using infrared imaging technology,can make up for the disadvantages of visible night monitoring,can block penetrate clouds and obstacles,realize all day long and long distance detection,large area coverage,the initial flame recognition and fire hazard(abnormal high temperature)problem in the field of recognition and real-time etc.in the forest fire prevention,is suitable for China’s national conditions,can be one of the important means of effective prevention of forest fires.Infrared image processing is an important part of infrared imaging forest fire prevention,which needs to be used for the detection of dim targets and the identification of the target with the shape of fire source in the near distance.Two.This paper makes a deep research on the technology of infrared imaging forest fire prevention:1.Image denoising,the infrared image characteristics of the target is unknown,the contrast is not high,the infrared image preprocessing prone image blur and edge information loss problem,proposed the use of simplified pulse coupled neural network method for infrared image denoising.Through the method of value improvement neural connections strength,the gray value of the pixel and neighborhood related to noise elimination;simplified calculation method of attenuation index threshold,which depends on the threshold magnitude,and can automatically obtain the best threshold value range.2.Image segmentation of infrared image preprocessing,after continued use of simplified pulse coupled neural network for target segmentation,in the neighborhood of the ignition state diagram use the iterative algorithm,threshold segmentation is constantly revised until after gray image cross entropy minimum value as the threshold for image segmentation.According to the experimental data,it can be seen that the use of PCNN network segmentation image has some advantages in forest fire prevention,and the segmentation effect is good,which is beneficial to image recognition.3.Build the small target sequence model.Aiming at the small target distance,no obvious shape feature,low contrast and low signal-to-noise ratio(SNR),the method ofsingle frame and sequence is used to detect small targets.Single frame intelligent detection based on adaptive morphological top hat transformation of structural elements to suppress background;on the basis of the target correlation with CFAR processing algorithm have a goal,to achieve filtering sequence detection by pipeline.4.To close the shape fire target segmentation,target feature extraction,by calculating the mutual information between the features,eliminate redundancy,the feature vector of the selected objective into the intelligent recognition device built by BP neural network classification.Using 20% of the sample data to identify the training,and then the remaining 80% data to identify the test,the output classification results.Based on the infrared imaging of forest fire research,implementation of small target detection and remote fire near shape target recognition,the validity of simulation results show that,it has certain theoretical significance and application value,which can be used in the infrared imaging of forest fire.