基于视频的火焰运动区域检测方法研究

Research on a Method of Detecting Flame Motion Region Based on Video

传统火灾检测方法主要是通过对火灾发生现场环境的温度、湿度、气体烟雾以及火焰辐射的紫、红外光,进行检测来发出警报,受空间距离的限制、环境因素变化的影响以及传感器稳定性的影响,误报情况严重。基于视频的火焰检测方法,非接触探测,不受空间距离、易爆、有毒等环境条件的影响,抗干扰能力强。通过数字通信和图像处理技术,视频火焰检测能够对火灾火焰可以进行时间和空间变化以及其他动态和静态参数的分析,处理速度快、准确率高,较好地克服了传统火焰检测方法的不足。该方法主要包括差分图像运动检测法、累积差分图像运动检测法、基于背景更新的运动检测法等几种。通过大量现场试验得出,基于背景更新的运动检测方法能够较准确地检测火焰的运动区域,误差较小。

The traditional methods of fire detection are mainly through detecting temperature,humidity,gas smoke and flame radiation purple infrared light in fire environment for alerting.The shortcomings are limited space distance,changes susceptible to environmental factors and the tranquilization of the sensor,seriously resulting in fase alarm.This method based on video uses the non-contact detection technology,not affected by the distance,explosive,toxic and other environment conditions and has a good anti-interference ability.It combines with digital communications and image processing technologies to analyze the flame characters of time and space and other dynamic and static parameters,having fast processing speed and high accuracy,better compensating the lack of traditional methods.This method includes differential image motion detection,cumulative differential image motion detection and motion detection based on background updating and others.Through many experiments,they find the optimal one is the motion detection based on background updating,which has the highest accuracy and the smallest error to detect flame motion region.