辽宁工程技术大学学报（自然科学版）2011，Vol.30Issue(5)：713-716,4.DOI:CNKI:21-1379/N.20111029.1325.006
煤矿井下图像型早期火灾探测
Early fire detection based on image in underground coal mine
摘要
Abstract
To overcome the shortcomings in traditional fire detection method in underground coal mine, a novel approach based on digital image processing and pattern recognition is proposed in this paper. In addition, the processes of infrared image obtaining, preprocessing and feature extraction are discussed in detail. According to the fire features at early stage, the fire information of multi-parameters is extracted from image sequence. Subsequently, the characteristics values are quantified and input into Support Vector Machine (SVM). The classifier is trained with SVM, and then the fire hazards and suspected objects are classified using the trained classifier. The experiment results demonstrate that the algorithm has high fire detection accuracy, lower false alarm rate and strong anti-interference ability. Also, the method is effective to small samples and nonlinear problems. This study is of significance to external fire forecast for coal mine.关键词
煤矿井下/早期探测/图像/特征提取/火灾识别/火灾信息/融合/SVMKey words
coal mine/ early detection/ image/ feature exaction/ fire recognition/fire information/ fusion/SVM分类
信息技术与安全科学引用本文复制引用
王媛彬,马宪民..煤矿井下图像型早期火灾探测 [J].辽宁工程技术大学学报（自然科学版）,2011,30(5):713-716,4.基金项目
国家自然科学基金资助项目(50977077) （50977077）
陕西省教育厅科研计划资助项目(11Jk0908) （11Jk0908）
西安科技大学培育基金(2010010) （2010010）
