›› 2015, Vol. 28 ›› Issue (8): 94-.

• 论文 • 上一篇    下一篇

一个基于视频图像检测火灾的算法

李红波,张其亮,杨杏华,宋英磊   

  1. (江苏科技大学 电子信息学院,江苏 镇江 212003)
  • 出版日期:2015-08-15 发布日期:2015-08-15
  • 通讯作者: 宋英磊(1977—),男,博士,教授。研究方向:算法设计,模式识别,机器学习等。E-mail:syinglei2013@163.com
  • 基金资助:

    江苏科技大学博士科研启动基金资助项目(635301202)

An Algorithm for Fire Detection Based on Video Image Analysis

LI Hongbo,ZHANG Qiliang,YANG Xinghua,SONG Yinglei   

  1. (School of Electronics and Information,Jiangsu University of Science and Technology,Zhenjiang 212003,China)
  • Online:2015-08-15 Published:2015-08-15

摘要:

针对传统火灾检测技术在面对大空间结构建筑及复杂环境中的不足,提出一种基于视频图像检测早期火灾的算法。该算法首先对视频图像进行阈值分割,然后根据火焰的颜色特征获取其中的疑似火焰区域。在此基础上计算疑似火焰区域的圆形度,并将圆形度与其整体特性相结合进行火灾的早期探测和预报。

关键词: 视频图像, 火灾检测, 火焰特征

Abstract:

Aiming at the shortage of traditional fire detection technology in large space structure building and complex environment,we develop a fire detection algorithm that can detect fire in early stage based on video image.The algorithm first uses threshold segmentation to segment the video image and obtained suspected flame regions based on the color characteristics of flame pixels.It then computes the circular degree of each suspected flame region and combines its circular degree with its overall characteristics to determine whether it contains flames or not.Testing results show that the algorithm can identify the flame regions from video data with a real-time fashion and can therefore be used for the early detection and prediction of fire.

Key words: video image;fire detection;characteristics of flame pixel

中图分类号: 

  • TP391.41