首页 | 本学科首页   官方微博 | 高级检索  
     检索      

低气压下火焰视频图像特征研究
引用本文:贾阳,林高华,徐高,王进军,方俊,张永明.低气压下火焰视频图像特征研究[J].火灾科学,2016,25(4):183-187.
作者姓名:贾阳  林高华  徐高  王进军  方俊  张永明
作者单位:西安邮电大学计算机学院,西安,710100;中国科学技术大学火灾科学国家重点实验室,合肥,230026;中国科学技术大学火灾科学国家重点实验室,合肥,230026;中国科学技术大学火灾科学国家重点实验室,合肥,230026;中国科学技术大学火灾科学国家重点实验室,合肥,230026;中国科学技术大学火灾科学国家重点实验室,合肥,230026;中国科学技术大学火灾科学国家重点实验室,合肥,230026
基金项目:国家自然科学基金联合基金(U1233102)、中央高校基本科研业务费专项资金(WK2320000032)及城市公共安全安徽省协同创新中心资助
摘    要:为了进行非密封飞机机舱内视频火灾探测技术的研究,借助中国科学技术大学QR0-12步入式环境低气压试验舱开展低气压下(100kPa,90kPa,70kPa,50kPa和30kPa)火焰视频图像特征研究。在实验舱中用正庚烷作为可燃物进行点火实验,拍摄火焰视频,研究低气压环境下火焰的颜色、空间变化、运动、相对稳定性、边缘粗糙度、相邻帧火焰区域面积变化率、面积重叠率、相关性特征。实验结果表明,火焰的颜色、空间变化特征不会随气压变化而变化;而火焰动态特征等都会因气压的不同而发生变化。因此,火焰的颜色和空间变化特征在低气压环境中仍可用于火焰区域分割和识别,而其他动态特征会随着气压发生变化,不能用常压下的方法来训练分类模型,但仍可以用以区别火焰和静止的疑似区域。

关 键 词:低压环境  正庚烷火焰  图像特征  视频火灾探测
收稿时间:2016/7/1 0:00:00
修稿时间:2016/8/25 0:00:00

Flame features under low atmospheric pressure based on video image analysis
JIA Yang,LIU Gaohu,XU Gao,WANG Jinjun,FANG Jun and ZHANG Yongming.Flame features under low atmospheric pressure based on video image analysis[J].Fire Safety Science,2016,25(4):183-187.
Authors:JIA Yang  LIU Gaohu  XU Gao  WANG Jinjun  FANG Jun and ZHANG Yongming
Abstract:In order to study the video-based fire detection technology in an unsealed aircraft cabin, experiments of flame combustion were conducted in a QR0-12 low air pressure cabin. The experimental pressure ranges from 30 to 100 kPa. N-heptane is used as the fuel in experiments. The examined features captured by CCD flame image include the flame color, spatial variation, motion, stability, and roughness. The change rate and overlap area and the correlation of the area of adjacent frames are calculated and analyzed. Experimental results show that flame color and spatial variance do not vary with the change of air pressure, and the dynamic flame features vary with the change of air pressure. Therefore, flame color and spatial variance can still be used in a low pressure environment for flame region segmentation and recognition. However, dynamic features vary with the air pressure, so the methods used under normal pressure cannot be used to train a classification model here. They can be used to distinguish flame and static suspected areas.
Keywords:Low air pressure  N-heptane flame  Image feature  Video-based fire detection
本文献已被 CNKI 等数据库收录!
点击此处可从《火灾科学》浏览原始摘要信息
点击此处可从《火灾科学》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号