Image Processing And Support Vector Machine Applied Research In The Field Of Fire Detection

The hazard of fire is tremendous, and the key to effectively prevent the occurrence of fire is the fire detection technology. In the current fire detection technologies, fire detection technology based on artificial neural network has been widely studied and applied. However, artificial neural network is vulnerable to the complexity of network and the sample. It has many defects such as over learning, lack of generalization, etc, resulting in false or missing alarms sometimes. Support vector machine has many advantages such as simple structure, global optimization, and good generalization ability. It can effectively avoid the emergence of the shortcomings of artificial neural network. It shows excellent performance in solving the problems of small sample, nonlinear and high dimensional pattern classification. Therefore, a fire detection technology based on Support Vector Machine (SVM) and digital image processing is proposed.The main results of the research are as follows:(1) Read a large number of the relevant domestic or literature about fire detection technology. Propose a fire detection technology based on Support Vector Machine and digital image processing to solve the shortcomings and deficiencies of traditional fire detection technology.(2) Systematically study the digital image processing technology and deal the fire images with the technology. Propose a suitable method based on significant color information for the fire image segmentation to extract a more complete suspected fire flame area. And the image preprocessing lays a solid foundation for feature extraction of the fire flame next.(3) In the analysis of a large number of fire flame pictures and common interference pictures, five representative features of fire flame are extracted, including shape of the fire flame, color of the fire flame, and texture of the fire flame, which effectively rule out the interference of common objects.(4) Systematically study the theory of SVM. Understand the principles of classification algorithm of SVM. Designed the SVM classifier for fire detection. The simulation shows that the fire detection technology based on SVM and digital image processing was effective and feasible. And this technology proved more excellent when it was compared with the fire detection technology based on BP neural network.