Research On The Fire Alarm System Based On Image Analysis

With the economic development and the rapid popularization of urbanization,the fires frequently occur in various living spaces both inside and outside the city.The traditional fire alarm system based on sensors such as temperature sensation and smoke sensor is prone to false alarm and leakage alarm due to the interference of environmental factors.The early fire detection and alarm system based on digital image processing technology uses high-definition cameras to monitor the scene and uses a variety of digital image processing algorithms to process the image sequence to extract the suspected area and then extract the static and dynamic features of the suspected area.At last,the pattern recognition technology is used to achieve fire detection.In comparison,the new automatic fire alarm system based on image processing technology can monitor and obtain more information,react and handle fire more quickly,and is more suitable for early fire monitoring and identification.In this paper,the current situation of fire alarm system at home and abroad is analyzed in depth.Then the related theories of image processing technology are studied in detail,including the image color model,gray processing,filtering,morphology and so on.Based on this,the fire image is segmented by the OTSU algorithm,and the flame target area is extracted by the background subtraction method.At the same time,according to the characteristics of the flame changes during the fire,the area features,morphological features,edge features,texture features and stroboscopic features are extracted respectively.In order to further eliminate the redundancy of feature information and reduce the dimension of the feature vector,the correlation analysis method is used to optimize the extracted features to obtain a more reasonable feature vector.Finally,based on the detailed analysis of the basic principle and implementation process of BP(Back Propagation)artificial neural network,the optimized feature vector is taken as the input of neural network.Through the adjustment of network parameters,the recognition of fire is realized,and the simulation shows that the method is correct and effective.Based on the research of image processing algorithms,this paper designs a wireless fire alarm system based on video surveillance technology,wireless bridge technology and VLC video development technology.Detailed analysis of the functional requirements of the system,and then gives the overall design of the system.At the same time,the design and implementation of wireless communication technology are discussed in detail.Finally,the VS2015 platform is used to design and develop the client software of the remote monitoring center,and a variety of image processing algorithms are incorporated into it to achieve the functions of fire detection and identification.