基于支持向量机的火灾探测技术  

Fire detection technology based on support vector machine

作　　者：张正荣[1,2] 李国刚[1,2] 

机构地区：[1]华侨大学信息科学与工程学院,福建泉州362021 [2]华侨大学嵌入式重点实验室,福建厦门361008

出　　处：《微型机与应用》2010年第24期70-72,76,共4页Microcomputer & Its Applications

基　　金：福建省自然科学基金资助项目(A0640005)

摘　　要：针对目前火灾探测方面的不足,提出了基于支持向量机的火灾探测技术。基于HSI颜色模型提取出火灾火焰疑似区域,在图像处理技术基础上获得早期火灾火焰的五个主要特征,采用支持向量机技术进行火灾识别。Matlab仿真实验证明,基于支持向量机的火灾探测技术识别率高,克服了神经网络过学习、容易陷入局部极小点等不足。该技术的研究在火灾探测领域具有重要的理论意义和实用价值。Concerning the current deficiencies of fire detection,a fire detection technology based on support vector machine was proposed.Firstly,the suspected area of fire flame was extracted based on HSI color model,then five main characteristics of early fire flame were obtained based on image processing technology,and use support vector machine technology for fire detection finally.Matlab simulation results showed that fire detection technology based on support vector machine had high recognition rate,it overcame the disadvantages of neural network such as over learning,being easily trapped in local minimum,etc.The technology has important theoretical and practical value in the field of fire detection.

关 键 词：图像分割 支持向量机 特征提取 火灾识别 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]