Research On Fire Situation Analysis Method Based On Infrared Image

Fire is a burning phenomenon.It will devour precious natural resources and pose a serious threat to national property and people’s lives.In order to detect the wild fires and to assess the current fire situation,a fire identification analysis method based on infrared images technology is proposed in this paper.The article research content includes image denoising,delimits suspected fire areas,extracts flame features,fire identification and analyze the fire environment.The data of infrared fire videos was taken at outdoor by my myself.The experimental results show that this algorithm has high accuracy and low miss rate.Focusing on the low SNR of infrared image,a new f-H filtering denoising algorithm combining edge information is proposed in this paper.The image can be changed by the weight value λ.On the one hand,the algorithm can reduce the image noise while preserving the edges,on the other hand,it can improve the SNR and resolution of the infrared image,the application effect is better than before.FCM algorithm is used to distinguish the target and background in the image,and the morphological closed operation is used to solve the problems of isolated noise points and small holes in the flame area,and finally we can get the suspected fire area in the infrared image.After this,The experiment extracts the feature of shape,texture,histogram feature and dynamic feature of the image flame.There are many features and they are interrelated.In order to improve efficiency,it is necessary to select the feature vector set.In this paper,KPCA algorithm is used to reduce unnecessary flame feature data,reduce the vector dimension,and obtain the optimal feature subset.In the aspect of fire recognition,the TWSVM classifier is realized through experiments,and a regular term is added to the matrix to eliminate the singularity of the matrix and improve the classification performance of the model.At the same time,considering the shortcomings of TWSVM in parameter selection,this paper proposes a QGA-TWSVM algorithm based on quantum genetic algorithm hyperparameter optimization,and the accuracy of fire flame recognition is 97.65%.In order to discover the recognition effects of different feature subsets,classifiers and parameter optimization algorithms,there are three experiments to found the conclusion.Finally,the article explores the influence of environment such as the wind direction and wind force of the fire site on fire,inferred the situation.