吉林大学学报(信息科学版) ›› 2019, Vol. 37 ›› Issue (6): 701-708.

• • 上一篇    

基于变分水平集图像分割的视频烟雾检测方法

李明明   

  1. 吉林大学大数据和网络管理中心,长春130012
  • 出版日期:2019-11-24 发布日期:2020-01-06
  • 作者简介:李明明( 1980— ) ,男,长春人,吉林大学高级工程师,硕士,主要从事计算机网络及应用、大数据和智能算法研究,(Tel) 86-13756550776( E-mail) lmm@ jlu. edu. cn。

Video Smoke Detection Method Based on Variational Level Set Image Segmentation

LI Mingming   

  1. Management Center of Big Data and Network,Jilin University,Changchun 130012,China
  • Online:2019-11-24 Published:2020-01-06

摘要: 为及时判断火灾发生情况,利于灭火工作,研究了火灾早期产生烟雾的情况下,通过固定监视探头捕捉
到的影像视频,针对视频中的火情烟雾进行检测,结合计算机视频检测技术的光流法和帧差法获得视频中烟雾
运动的区域,基于某一时刻摄像机不动的情况下得到两帧连续的图像并检测视频中的运动区域,提出了一种不
需初始化的变分水平集算法。利用该算法对烟雾的视频进行分割,将运动区域从背景中分割出来,对烟雾进行
检测分析。从而使通过烟雾判断火情更精确,检测结果更实时,提高了视频指挥灭火工作的效率。

关键词: 视频检测, 变分水平集, 光流法, 帧差法, Chan-Vese 模型, 视频分割

Abstract: In the case of smoke generated in the early stage of the fire,the video captured by the fixed
monitoring probe is used to detect the fire smoke,and the frame difference method of the computer video
detection technology is used to obtain the area of the smoke movement. When the camera is not moving,two
consecutive images are obtained and the motion region in the video is detected. The video of the smoke is
segmented by the variational level set method,and the motion region is segmented from the background to detect
and analyze the smoke. Therefore,it is more accurate to judge the fire through the smoke,and the detection
result is more real-time,which improves the efficiency of the video command fire fighting work.

Key words: video detection, variational level set, optical flow method, frame difference method, Chan-Vese model, video segmentation

中图分类号: 

  • TP311. 5